Revisiting Dividend Growth Predictability via Dividend Yield

Abstract

One of the main theoretical implications of the present value approach on firm valuation is the hypothesis that dividend yield has a predictive power on future dividend growth. The relevant literature, however, was not able to provide evidence that clearly supports this hypothesis. In this paper we cope with the two main reasons that raise the econometric complexity on testing the dividend growth predictability hypothesis, namely, the seasonality effects that appear when higher frequency data are used, and the effect of price volatility on the computation of dividend yield. Specifically, an application of a the Mixed Data Sampling (MiDaS) technique allows us to use monthly dividend data in order to test the hypothesis that dividend yield explains the future annual dividend growth. In order to cancel out the effects of price variation on dividend yield we use a smoothing technique, and we identify the component of the smoothed dividend yield that offers predictive power. Empirical evidence from US, UK, Canada, Germany, France and Japan, strongly supports the dividend growth predictability hypothesis.

Keywords: Dividend growth, dividend yield, predictability, dividend smoothing, mixed frequency data analysis

JEL Classification: C53, G14, G15, E44, E47