Volatility, Earnings, and Multiples

Abstract

The author decomposes the variance of stock returns into a fundamental component, proxied by the variance of their earnings; a behavioral component, proxied by the price-to-earnings ratios’ variance; and a joint component. By sorting stocks according to each of the three components, he found the usual low-volatility pattern, whereby stocks with low variance outperform high variance ones on an absolute as well as a risk-adjusted basis. The evidence confirms the findings in existing literature that both behavioral and fundamental reasons lead to the volatility effect. He further investigates the mutual effect of each component, by applying standard double-sorting techniques to neutralize one component while building sorted portfolios according to the other component. When considering the behavioral effect, the cross-sectional differences in the fundamental component disappear: There is almost no difference between low- and high-variance stocks. On the contrary, even when neutralizing the fundamental component and sorting according to the behavioral component, he still finds evidence of the volatility effect. The results do not rule out fundamental reasons behind the volatility effect. They instead stress the pervasive effect of the behavioral component, which remains significant even after accounting for the fundamental component.

TOPICS: Security analysis and valuation, fundamental equity analysis, performance measurement

Key Findings

• ▪ Stocks with low variance outperform stocks with high variance on a risk-adjusted basis. This so-called “low-volatility” anomaly is still measurable if one looks at either stocks’ earnings variance or stocks’ price-to-earnings ratio (P/E) variance.

• ▪ When accounting for the price-to-earnings volatility, however, stocks with low variance do not outperform their high-variance peers. On the contrary, even by accounting for earnings variance, stocks with low P/Es still outperform their high-variance peers.

• ▪ Part of the well-known volatility effect therefore seems to be related to the variance in price-to-return earnings.