Forecasting dividend growth: The role of adjusted earnings yield

Abstract

This paper revisits the predictability of dividend growth. Motivated by the dividend partial adjustment model, we decompose the earnings yield into smoothing and residual components. The residual component reflects the variations in the forecast of dividend dynamics and thus forms a powerful predictor of dividend growth. Empirically, the proposed predictor shows significant in-sample predictive power for aggregate dividend growth at both monthly and annual frequencies over several forecast horizons. The regression results are robust to dividend reinvestment strategies and economic status. More importantly, the proposed predictor contains significant out-of-sample predictability, outperforming the historical mean benchmark and a list of popularly used financial and macroeconomic predictors in the literature.