Time-Varying Predictability for Stock Returns, Dividend Growth and Consumption Growth

Abstract

Using a state-space model, this paper examines time variation in the predictive regressions for stock returns, dividend growth and consumption growth. Moreover, we linked time variation explicitly to movements in economic factors that can account for risk and cash flow. Results support the view that stock return predictability is enhanced when risk is high (negative growth, higher volatility and positive growth/return covariance). In contrast, dividend growth and consumption growth predictability is enhanced during economic expansions. These results are supported by subsample analysis and a vector autoregressive approach. Furthermore, these latter exercises may uncover differences in the stock return predictability relationship when viewed over different time horizons. Overall, the paper contributes to the literature by highlighting the different nature of returns predictability, which arises largely through the risk channel, and dividend and consumption growth predictability, which arise through the cash flow channel.