Center of Asset and Wealth Management Blog

Predicting returns: Do not trust dividend yields in small industries

Long time, dividend yields have proven to be a proxy for returns, but the general validity of this statement is questionable.

© Adobe Stock/Funtap

Long-term stock returns are predictable. The existing literature mostly confirms this statement, especially in the US stock market in post-war years. In practice, however, investors do not always hold well-diversified market portfolios. Instead, investors often pursue an industry rotation strategy and/or concentrate their holdings in sectors in which they have more experience or private advantages. Further explanations for this approach can be behavioral biases, financial constraints, and limited borrowing opportunities. At the same time, investments in certain industries can provide substantial profits to investors. Thus, addressing return predictability across US industries is important for investors.

In her recent paper with the title “Stock Return Predictability: Evidence Across US Industries”, Quynh Pham investigates two questions.

  1. Do the long-term return predictive patterns hold across U.S. industries?
  2. Which hypothesis can explain the different predictive patterns, if any?

Empirical results illustrate two important findings.


First, stock return predictability differs across US industries.

Return predictability patterns do not always hold across US sectors. In fact, a predictability heterogeneity exists over industries. Stock returns tend to be predictable in sectors with large firms or in large sectors such as, tobacco, communication, and petroleum & natural gas, and less so in small sectors, such as coal, steel, and construction materials. The subsequent figure shows that the long-run return predictive coefficients tend to increase along with an increase in the average firm size within industries. Moreover, the correlation between the two variables is 0.51 and statistically significant, confirming that industries with large firms have more likely predictable returns.

Second, large industries tend to smooth their dividend payout.

Why are stock returns highly predictable in large industries but less so in smaller sectors? In light of the dividend-smoothing hypothesis, there is supporting evidence for the higher dividend smoothness in large industries than in small ones. A high degree of dividend smoothness disconnects the link between dividend yield fluctuations and future dividends, which lessens dividend growth predictability and enhances return predictability. Consequently, stock returns are more predictable in large sectors and less so in small sectors.

Large industries tend to smooth their dividend payout, which makes their dividend yields more persistent. Therefore, dividend yields might be good proxies for predicting long-term returns in industries with large firms but not for industries with small firms.


Please refer to the original paper for further details:

Pham, Q. T. T. (2020). Stock Return Predictability: Evidence Across US Industries. Finance Research Letters, in press.