Does the dividend/price ratio predict returns?

We discuss the predictability of aggregate stock market returns using currently observed dividend-price ratios

Return predictability is of considerable interest to practitioners who can develop market-timing portfolio strategies that exploit predictability to enhance profits.

The question of today is whether we can use the currently observed dividend/price ratio to predict market returns. Specifically, by dividend/price ratio we mean the weighted-average price-ratio of the market, and by market returns, we mean, the next period (quarter? year?) weighted-average returns of the market. So, our question is really on aggregate predictability, not on cross-sectional predictability.

Following an old approach set out by Campbell and Shiller (1988), it can be shown that today’s dividend-price ratio is mechanically related to next period’s returns and dividend growth. In other words, we should be able to predict next period’s returns and dividend growth by looking at the current dividend-price ratio.

Let’s look at the evidence. The table below is from Koijen and Van Nieuwerburgh (2011) and it reports the results of a series of regressions in which respectively returns (Panel A) and dividend growth (Panel B) are regressed on the previous year dividend-price ratio. The parameters of interest are kr and kd, which respectively capture the coefficients of the dividend-price ratio in the returns and dividend growth regressions.

In the benchmark period 1926-2009, for the case in which dividends are reinvested at Rf (top-left panel), kr is positive (0.077) but statistically insignificant. For the post-war period 1945-2009, the coefficient is also positive (0.130) and statistically significant. Things look better if one assumes that dividends are reinvested at Rm (top-right panel). Still, the R2 of the regressions is at best 10.84%.

Is this much or little? Campbell and Thomson (2008) argue that given an average Sharpe ratio of 36% in the US market, an R2 of 10.8% implies that a dynamic strategy with market-timing based on the regressions reported above, leads to a doubling of returns, with respect to a static buy-and-hold strategy. So maybe 10.8% is not too little after all…

Panel B shows that the predictability of dividend growth is also quite poor. The coefficient kd is never significant.

What do we learn?

You can read this has a glass half-full or half-empty. On the plus side, aggregate returns are somewhat predictable using the dividend-price ratio. Albeit limited, such predictability can lead to superior returns via active index-investing, for example, using both long and short ETFs on the S&P500. On the down side, a 10% R2 seems rather low, particularly if not matched by a strong R2 on the dividend growth regressions.

What is clear is that venturing into speculative claims that markets are over-valued or under-valued on the basis of dividend-price ratios is broadly-speaking a futile exercise.

References

1. Koijen, R.S. and Van Nieuwerburgh, S., 2011. Predictability of returns and cash flows. Annu. Rev. Financ. Econ., 3 (1), pp.467-491.

2. Campbell, J.Y. and Shiller, R.J., 1988. The dividend-price ratio and expectations of future dividends and discount factors. The Review of Financial Studies, 1(3), pp.195-228.