Minimum volatility: what's in a name?
by Ehren Stanhope, CFA, Factor Investor
"A great ad campaign will make a bad product fail faster. It will get more people to know it's bad." - Bill Bernbach, DDB
Mad Men watchers may recognize the name Bernbach from the quote above. Bernbach is referred to in the second season as the innovative competitor firm that challenges Sterling Cooper's orthodoxy. Bill Bernbach was the brain behind several successful campaigns, including Avis' We Try Harder and Volkswagen's Think Small. The quote couldn't be more applicable to many of the Smart Beta options out there. There is likely nothing cataclysmic on the horizon with these investment vehicles, but investors should be aware that they are generally repackaged versions of existing strategies in diluted form.
In my last two posts, I explored a fad which has washed over the investment community--low volatility investing. In my first post, I looked at the historical performance oflow volatility stocks and identified that outperformance from the factor tends not to be very consistent over time, but is instead clustered in certain periods. 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. I then established in my last post that Low Volatility investing is likely just value investing in re-packaged, and expertly marketed, form (hence the Bernbach quote above). This is best captured by looking at hypothetical portfolios of stocks ranked by volatility, below, and then determining the average PE Ratio.
The chart above shows a clear relationship between the average PE over the last five decades, and the valuation of the portfolio--the higher the volatility, the higher the valuation. I have no beef with Low Vol portfolios in the same way that I have no beef with an investment manager who chooses to run his portfolios differently than I do. However, I think it's important to understand what you own. And behind all the whiz-bang advertising, I think the expectations of most Low Vol investors are mis-guided. And so as promised, I dive into the characteristics of a popular Low Vol ETF that has garnered $12 billion in assets in five years.
iShares MSCI Minimum Volatility ETF (USMV)
USMV "seeks exposure to U.S. stocks with potentially less risk", per its website. Risk is a funny word. It means many things to many people. Here, risk is defined as volatility. Absent the intervention of the iShares legal team, I suspect the ETF's name would have been named "Minimum Risk ETF". Who wouldn't want that? But when you look under the hood, you find that there is lots of risk in the holdings, even by the ETF's own definition of risk--volatility. To be fair, the ETF seeks a low volatility portfolio, which means that not every constituent stock needs to be low volatility on a standalone basis. One would expect a name or two to exhibit higher volatility, but would you expect a stock owned by a "minimum volatility" ETF to have a 12-month standard deviation (volatility) of 47.5% per year?
It turns out actually, that USMV holds many names that would likely qualify as "high" volatility. From 1964-2015 the average 12-month standard deviation (volatility) for stocks falling into the High Vol bucket was 27.0% annualized. Clearly, the stock with a 47.5% standard deviation mentioned above, is to the far right on the spectrum below.
So how does this play out in the portfolio? Here I have to pause for a moment and give credit for the author of the Investor's Field Guide for inspiring the bubble charts below. Bill Bernbach would be proud; a picture is worth a thousand blog posts.
For the chart below, each bubble represents a holding in the USMV ETF as of April 8th, 2016. The bigger the bubble, the greater the weight in the ETF. Market cap is measured on the vertical y-axis using a log transformation. The higher on the y-axis, the larger the stock. You'll notice a bias towards larger cap stocks in the ETF. A bubble's position on the x-axis is dictated by its trailing 12-month volatility. The lower the volatility, the further to the left on the chart. The higher the volatility, the further to the right.
The first thing you notice is that there are a lot of bubbles to the right of 20% and 30% volatility. There are two curious outliers that have hefty weights, including the nefarious one mentioned earlier with 47.5% annualized volatility--Newmont Mining. Other outliers include Facebook, another large weight in the portfolio, and Cabot Oil & Gas, a much smaller weight. It is certainly the case that there are a number of Low Vol stocks in the ETF, but there are also a number of High Vol stocks. How is that?
It turns out that the ETF doesn't just invest in the lowest volatility stocks. It actually uses an algorithmic framework which seeks to minimize volatility for the total portfolio not at the individual name level. There are some advantages and disadvantages to this. Unfortunately, the advantages fall mostly to the ETF sponsor. I'll explain...
Advantage: Co-movement = larger scale
Whenever you tilt the allocations in a portfolio away from traditional market cap weights, you constrict the capacity of the product. My colleague Patrick O'Shaughnessy has written about this at length. In our case, USMV is tilting towards low volatility stocks, but there are only so many to go around. The solution is to look to stock comovements to optimize at the portfolio level for minimum volatility. In this manner you can include a stock like Facebook with greater than 30% volatility in the portfolio if it has low or negative correlation with another portfolio holding like say, Berkshire Hathaway. The volatilities of the two lowly correlated names can effectively cancel each other out. At $330 billion in market cap, adding a name like Facebook to the portfolio significantly increases the overall capacity of the ETF.
Disadvantage 1: Co-movement = Correlation = Constantly Changing
Correlations in markets are key. It is what underlies the principle of diversification of financial assets. For example, we invest in stocks and bonds because bonds tend to do well when stocks are doing poorly, and vice versa. They act as ballasts to each other. This principal extends to stock-only portfolios as well. The problem, as anyone who owned a stock portfolio in 2008 observed, is that when indiscriminate sell-offs occur, EVERYTHING goes down, at the same time. Historical correlations breakdown. Allocations based on minimizing historical volatility become irrelevant. There is a simple solution to this problem--rebalance the portfolio based on new information. Unfortunately, ETF's like USMV are based on "indices" that only rebalance twice a year--the last day of May and November. They are slow to adjust to changes in correlations.
Disadvantage 2: Volatility is not a panacea
A portfolio can be viewed as a series of exposures that need to be controlled. Theoretically, you want to make your bets based on some inference, but control for any other exposures where you do not have an edge. This begs the question: what other exposures are we taking on in USMV for the benefit of a low volatility portfolio?
The chart below is similar to the one above except that Quality is represented on the vertical y-axis. What becomes readily apparent is that the ETF's Quality exposures have no real correlation with low volatility. The exposures in the portfolio are broadly dispersed vertically from low Quality (percentile score of 100, higher on vertical axis) to high Quality (percentile score of 1, lower on vertical axis). Here again, I highlight the same outliers by volatility. Facebook is not only highly volatile, but is also of relatively poor Quality.
The chart below swaps out Quality for Value on the vertical axis. Here again, you see wide dispersion vertically. Intuitively, you would expect the bubbles to be concentrated in the bottom left of both the chart above and below. If anything, on the Value spectrum below, the bubbles appear to be concentrated in the 50th to 90th percentile, suggesting the majority of the portfolio is more expensive than 50-90% of Large Stocks on the market.
On the basis of Value score, our outliers also fall into expensive camps in the market--a Value Percentile score of 100 , i.e. Facebook in this case, is more expensive than almost 100% of the market. Again, intuitively we would expect the bubbles to converge in the bottom left based on the first chart in this post, which shows Low Vol as historically associated with cheapness.
For fun, I'll combine the two charts above and plot Quality and Value scores together on the vertical and horizontal axes, respectively. Who would want to own this mish-mash of low Quality and expensive Value, which seems to be drifting to more expensive and lower quality portion of the chart (upper right).
Rising Tides Lift All Ships
It turns out that Low Vol has become a fad. In one year as of March 2016, USMV's assets increased $6.6 billion to $11.3 billion in assets--142% growth in assets! And with that, valuations are-a-rising on almost any metric you look at. The weighted average PE ratio by my calculations was 33.0x as of year end 2015. Below, I've included a historical time series since the ETF's inception for a wide array of value metrics. Whether you are looking at cash flows, sales, earnings, or total yield, USMV has become increasingly more expensive. Based on the last two charts which look to the quality of underlying holdings, Earnings Quality and Financial Strength have been gradually deteriorating since late 2014 (the higher the score, the worse the ranking).
Caveat Emptor
If the long-term benefits of Low Vol investing are actually associated with some sort of Value premium, then current Low Vol investors should beware. Valuation for Low Vol stocks is expensive. A once good idea has run its course. All of the research I have done or reviewed suggests that high current valuation leads to low future return. Valuations on Low Vol have risen substantially in just the last few years. Based on the PE ratio, USMV is 80% more expensive than it was when it launched. In a best case scenario, Low Vol seems destined for mediocre future returns. In a worst case, the correlations and volatilities underlying the ETF's holdings could change quickly, providing an unexpected return profile for investors whose expectations have been managed for downside protection.
Data Notes: All references to volatility are the trailing 12-month annualized standard deviation of return. Value is a multi-factor ranking based on sales, cash flows and earnings. Financial Strength represents a ranking based on balance sheet strength. Earnings Quality is a ranking based on the conservatism of a company's earnings.
Copyright © Factor Investor
