Volatility is a value factor

In my previous post, I looked at the historical performance of investing in low volatility stocks and identified that outperformance from the factor tends not to be very consistent over time, but is instead clustered. That raised some questions on whether volatility is a true investment factor, or if it's positive benefits are the product of other, more robust, investment factors.

Below again are the returns for various portfolios ranked by volatility (12-month standard deviation). We find that Low Vol portfolios do outperform, and High Vol portfolios do underperform. However, when we dug further into the consistency of performance, Low Vol was particularly lacking, and it was actually the middle deciles (4 and 5) that had the most consistent outperformance on a rolling 5-year basis.

This elicited some questions on the true driver of the volatility effect. The chart below looks at the average Price-to-Earnings Ratio (PE) of each of the Volatility deciles from 1964-2015. Here, a clear trend tied to valuation emerges. The inference is that any advantages associated with low volatility investing could be fully explained by our good ole' friend Valuation.

Luckily, we can run some simple tests via regression to test this hypothesis. If the hypothesis turns out to be true, then Low Volatility investing may just be value investing in re-packaged, and expertly marketed, form. Below are the regression results for the Low Vol and High Vol deciles (see notes at the end of this post for the regression parameters ). I extend the four factor Carhart framework to include some quality themes that we have found to be robust in our research.

The results are quite revealing. The top green box includes the coefficients for the Low Vol decile. Note first off that the coefficient for the market factor (Mkt) is 0.78. This is the Beta of the strategy versus an equal-weighted Large Stocks, which as we would expect, is much lower than the overall market. On the other end of the spectrum, the High Vol decile at the bottom of the panel, sports a Beta of 1.18--significantly higher than the overall market. The negative Size coefficient of -0.12 suggests Low Vol is typically associated with larger firms. High Vol has a positive Size coefficient of 0.20, which indicates smaller firms typically have higher volatility.

After the market factor, the variable with the greatest statistical significance is Value. The Low Vol decile has a 0.16 exposure to Value, which indicates Low Vol does in fact have a value orientation. The High Vol decile has the opposite exposure. It's coefficient of -0.52 suggesting it is heavily associated with expensive stocks. The only other theme that is statistically significant for Low Vol is Financial Strength. Low Vol has a positive exposure, which indicates the stocks in this bucket tend to have stronger balance sheets. The High Vol group does not have statistically significant exposure to Financial Strength, but it does have positive and significant exposure to Momentum. Given that the High Vol decile has such a high Beta of 1.18, it makes sense that those stocks would also have positive exposure to the Momentum factor. Markets go up about 70% of the time in rolling 12-month periods, so any strategy with strong Beta exposure should have positive Momentum exposure as well.

This information seems to confirm that the Low Volatility craze of the last few years is a bit misguided. Low Vol is not a panacea. Controlling for exposures to Value, Size, and Financial Strength, Low Vol's intercept (regression speak for alpha) effectively falls to 0. Since Low Vol investing simply tilts towards large firms that are priced at a discount and have strong balance sheets, we should be able to assess the potential for future outperformance based on current valuation levels.

On this aspect, buyer beware. At a PE ratio of 19.0x trailing earnings, Low Vol is currently 34% more expensive than its 52-year historical average, and is right on par with the 19.0x PE ratio of the S&P 500 index. In a recent piece, Rob Arnott, refers to a "Smart beta crash". While it would be impossible to predict such an event, it seems likely that Low Vol stocks are destined to add low return to their low volatility characteristic moving forward.

In the next post, I'll relate the info above and in my first post on Low Vol to the dominant ETF's which have garnered $16 billion in assets since their launch in 2011.



Regression notes: The regression was run using an R linear regression package. The Mkt variable is an equal-weighted universe of stocks (Large Stocks) traded on the NYSE, AMEX and NASDAQ that have a market cap greater than the inflation-adjusted average. Utilities are excluded. The remaining variables represent the excess return, or spread, generated from the high minus low (HmL) decile for each variable. Value is a multi-factor ranking based on sales, cash flows and earnings. Momentum is 6-month momentum. FinStr represents a ranking based on balance sheet strength. EarnQual is a ranking based on the conservatism of a company's earnings. EarnGro is a ranking based on profitability and earnings growth.