The Case for High-Volatility Stocks
The surprising power of the short interest signal.
More Is More
Low-volatility investing is fashionable. The tactic has spread from its original field of U.S. equities to foreign stocks, bonds, and other asset classes. Morningstar now tracks 14 exchange-traded funds that follow low-volatility strategies.
Low-volatility investing entails buying the tamest securities within a given investment sector. The logic is that investors are risk-averse when conducting asset allocation. Once their allocation is set, though, investors tend to prefer risk, as they seek to maximize returns for each asset class, and (per standard investment theory) they figure that the securities with relatively high volatility have relatively higher expected returns. They prefer risk so much that they overpay for higher-volatility issues. Which makes the neglected, seemingly dull securities the better investment bet.
Two professors offer a new take on this narrative. In The Long and Short of the Vol Anomaly (jargon alert!), Bradford Jordan and Timothy Riley examine U.S. stock-market returns from 1991 to 2012 and affirm that, yes, low-volatility stocks did outgain their higher-volatility peers.
But wait, there's more. It turns out that high-volatility stocks, like the little girl with the curl, have two moods. They are either very, very good--better than low-volatility stocks, in fact--or they are horrid.
The mood's indicator is the level of short interest. The authors find short sellers to excellent forecasters. With both low- and high-volatility stocks, securities that have relatively low levels of short interest* outgain securities that have high levels of short interest. The effect is modest with low-volatility stocks but not so with high-volatility securities, where short interest carries a very strong signal. It's so strong, in fact, that high-volatility stocks with low levels of short interest gain the most.
(*The authors measure the level of short interest as the raw dollar amount of short interest in a stock divided by that stock's average daily trading volume. This calculation is called "days to cover.")
(A technical point: The standard deviation of both high-volatility groups in the second table is higher than the standard deviation of the single high-volatility group in the first table. This would not make sense if the tables showed the average risk of each stock in the groups, because how can both halves be higher than the whole? But it is possible when the calculation is done by measuring the standard deviation of portfolios formed from those groups of stocks, which is what the authors did, following academic convention.)
Here are the data in pictures.
The first graph shows low-volatility stocks plotted against the entire stock market (as measured by the Center for Research in Security Prices) and against high-volatility stocks. Sure enough, low-volatility stocks finish far ahead.
The second graph shows the four quadrants of low volatility/low interest, low volatility/high interest, high volatility/low interest, and high volatility/high interest. Note the massive disparity between the two high volatility portfolios--the jagged high volatility/low interest blue line that eventually lands on top, and the sad high volatility/high interest dotted red line that hugs the x-axis. Like purchasing a vineyard, the tactic of buying risky stocks that are disliked by short investors turns large fortunes into small ones.
The authors write that their evidence indicates that "the current 'low vol' investing fad has little or no real foundation." I'm not sure why they believe that. After all, by their own numbers, low-volatility stocks have posted higher returns than the rest of the stock market, and (duh) with lower risk, too. And while both flavors of low-volatility investing, low and high short interest, do trail the high volatility/low interest quadrant for total returns, they surely exceed on digestible performance. Even the risk-tolerant will shudder at the shape of that blue line.
These findings seem at first to be driven by the momentum effect. The difference in momentum between the two groups of stocks, high volatility/low interest and high volatility/high interest, is massive. The median trailing 12-month return for the high volatility/low interest stocks (equally weighted) is 10.5%, while the return for the low volatility/high interest stocks is negative 14%. Twenty-five percentage points! On over 40,000 observations! That, friends, is a pattern.
However, the tale is rather more complicated. For one, the authors find an annualized alpha of 6.6% for the asset-weighted version of the high volatility/low interest portfolio, regressed against the Fama-French-Carhart model that includes momentum as one of its four factors. For another, within the high volatility/low interest group, the best-performing stocks had high negative momentum. That is a very large puzzler--as well as an indication that momentum is far from the sole answer.
The story as I read it:
I'm not sure how much this study directly affects investors. Low-volatility strategies remain broadly appealing. Returns figure to be at least competitive--not only in U.S. stocks, which this paper studied, but in other asset classes as well--and the ride will be smooth. In contrast, a high volatility/low interest strategy looks to be only for specialized tastes. Even if the stocks perform according to expectations, the ride will be terrifying.
However, the power of the short interest signal is noteworthy. That indicator could be used in a variety of ways, whether as an additional tool to be used for selecting stocks or as a factor used by an ETF. Also, the authors certainly succeeded in demonstrating that there are high-volatility stocks and then there are high-volatility stocks. There's no reason to shun the breed as a whole.
John Rekenthaler has been researching the fund industry since 1988. He is now a columnist for Morningstar.com and a member of Morningstar's investment research department. John is quick to point out that while Morningstar typically agrees with the views of the Rekenthaler Report, his views are his own.