Dividend Momentum and Stock Return Predictability: A Bayesian Approach

A long tradition in macro-finance studies the joint dynamics of aggregate stock returns and dividends using vector autoregressions (VARs), imposing cross-equation restrictions implied by Campbell-Shiller (CS) identity to sharpen inference. We take a Bayesian perspective develop methods draw from any posterior distribution VAR that encodes priori skepticism about large amounts return predictability while CS restrictions. In doing so, we show how common empirical practice omitting dividend growth system extra restriction is not persistent. highlight persistence induces previously overlooked channel for predictability, which label ``dividend momentum.'' Compared estimation based on OLS, our restricted informative prior leads much more moderate, but still significant, degree with forecasts are helpful out-of-sample realistic asset allocation prescriptions Sharpe ratios out-perform benchmarks.