Volatility is not necessarily the investor’s best friend. At least not all of it. Volatility to the upside is desirable; but not volatility to the downside. Still, as the saying goes, there’s no free lunch. Investors willing to take downside volatility risk should be rewarded. Hence the risk premium for stocks as against bonds and cash.
Still, risk is a long-term game. And not all risks are rewarded commensurately. This surfaces in the “low-volatility anomaly.” Portfolios of low-volatility stocks, both in the U.S. and in international markets, have higher returns than portfolios of high-volatility stocks, thus contradicting a fundamental tenet of investing. A paper by Rodney Sullivan of the CFA Institute and Xi Li of Boston College entitled “Why Low-Volatility Stocks Outperform: Market Evidence on Systematic Risk Versus Mispricing,” lays out the stakes.
“Should the anomaly be related to systematic risk, then the excess returns can be viewed as arising from some, as of yet unknown, common risk factor(s). Alternatively, it may be driven by a mispricing, as perhaps associated with an imperfection such as investor irrationality connected with volatility.”
There have been a couple of attempts to explain the anomaly systematically. One could be constraints on margin loans for high-volatility stocks. The global co-movement of low-volatility stocks is also suggestive of a systematic risk factor, rather than an aggregation of idiosyncratic or stock-specific risks.
Then again, it could be price. For example, “the efficacy of trading the low-volatility factor is very limited due to high transactions costs directly associated with attempting to extract the anomalous excess returns.” Arbitrage is not costless.
To get at whether systematic risk is at work, Sullivan and Li use factor loadings and firm-specific characteristics. More specifically, they write: “A particular factor loading provides an estimate of that factor’s risk premium. Thus, when considering the low volatility anomaly, for the systematic risk explanation to be valid, those stocks with a low factor loading on the low-volatility factor would necessarily have higher stock returns as compared to those stocks with a high factor loading. This pattern should be observed irrespective of the absolute level of stock volatility. If however, after controlling for the observed level of return variability, loadings on the low-volatility factor are unable to explain cross-sectional stock returns, then we can reasonably conclude that the low-volatility anomaly is consistent with some market mispricing.”
They do their analysis using the Fama-French factors of book to market, market cap and CAPM beta, or excess returns over T-bills, As a factor in its own right, idiosyncratic volatility (IVOL) then becomes the standard deviation of the residual return.
But is idiosyncratic risk a factor, or a firm-specific characteristic? Here Sullivan and Li are treading on an old methodological debate. How do you define stocks? Do you group them by their surface characteristics, such as value and low volatility? Or do you instead pre-sort them, by using such factors as value and low volatility to look at their co-variances? Another way of putting it is whether value stocks behave similarly because there is an underlying common risk premium, which suggests a systematic effect, or whether they behave similarly because they are under regarded and thus mispriced: a firm that loads high on a distress factor may not, in fact, be distressed.
Sullivan and Li try to take both explanations into account. “When estimated, our model will load most heavily on those risk factors potentially responsible for the return predicting powers of the IVOL characteristic (if risk is indeed the driver). This procedure extracts risk factors even if the researcher does not directly observe the factor structure underlying stock returns.”
They sort stocks by factor loadings, and also by controlling for factors, to isolate low volatility. In the tests, it is the idiosyncratic characteristic, not the factor loading, that is predictive of future returns. That argues against a systematic risk factor somehow vectored by low-volatility and in favour of a mispricing error: that investors shun low-volatility stocks. But the authors remain agnostic on what the source of that error may be.
And yet, it sheds light on common wisdom: like broccoli, low volatility is good for you.