Low Volatility ETFs and Their Underlying Indexes – USMV, EFAV, and SPLV

Weighing between risk and return, most investors actually would go for risk aversion if they have to take a side. Such preference is satisfied by constructing minimum-variance and managed volatility indexes and corresponding ETF products were launched to the market early in the 1990s. The top three – USMV, EFA, and SPLV – were at $12.7, $6.7 and $6.5 billion AuM as of the middle of 2017, are now (11/28/2018) soared to $18.14, $8.9, and $8.14 billion respectively.

iShares Edge MSCI Min Vol USA ETF, USMV, tracks the MSCI USA Minimum Volatility Index. The MSCI Parent Indices serve as the universe of eligible securities for optimization, which is based on the previous Barra Global Equity Model (GEM). The MSCI Minimum Volatility Index seeks to have the lowest absolute volatility based on a set of constraints using the most recent release of the Barra Open Optimizer, which determines the optimal solution, i.e. the portfolio with the lowest total risk, using an estimated security covariance matrix under the applicable investment constraints.

EFAV, The iShares Edge MSCI Min Vol EAFE ETF, tracks MSCI EAFE Minimum Volatility Index, which is created with the same methodology but for an international universe.

SPLV, the PowerShares S&P 500 Low Volatility Portfolio is built upon S&P 500 Low Volatility Index, which is not focused on the low volatility of the whole portfolio, but on picking individual stocks with low price fluctuation over years. The selection of index constituents is done as follows:

  1. Using available price return data for the trailing one year of trading days leading up to each index rebalancing reference date, the volatilities of the constituents within each eligible universe are calculated.
  2. Constituents meeting eligibility requirements as described under Eligibility Criteria are, then, ranked in descending order based on the inverse of the realized volatility. The top securities with the least volatility, as determined by each index’s targeted constituent count of the index.

Comparing these three ETFs’ past price performance, both USMV and SPLV well track the performance of S&P500 broad U.S. market, while, as expected, EFAV’s return line divert quite significantly because of its international scope. It seems in EAFE market, low volatility stocks yield less compared to their counterparts in the U.S. market.

Apparently, USMV has been consistently out-performing the competitor low vol ETFs. In addition, its expense ratio is low at 0.15%, compared to it immediate follower SPLV’s 0.25%. It’s not surprising that USMV leads and makes the most rapid growth in terms of fund in-flow in the last year. 

Now let’s take a look at the fluctuations of their prices over the last 6 years  to ascertain which one is truly low volatile. We take the price point at each month end, calculate their average and standard deviation for each ETF, and then compute each month’s Z-score accordingly, charted below. We can see that both USMV and SPLV are quite tightly tangled and both are slightly better than SPY, especially on drawdown moments. 

Combining performance and volatility to compare USMV and SPLV, one can draw the conclusion favoring USMV. Now since the two underlying indexes take very different approach with SPLV’s straightforward, while USMV’s seems more abstract, relying on its Barra risk model and optimizer. We’d dig deeper into them to see if FactSet’s cognitivity risk model could offer a rival or superior index product.

 According to their methodology, I quoted the detailed construction steps:

 With the arsenal of Barra Equity Model and Barra Open Optimizer, it’s not tricky except computationally intensive to try out various weight combinations. And then apply the classic  Markowitz’ minimum variance model to find the portfolio residing at the left-bottom corner – lowest risk relative to optimal, acceptable return. 

It seems Barra’s risk models are not particularly good at addressing fat tail risk, I think cognity risk model could do a better job given the existing risk factors are all captured and optimizer is technically universal once the parameters are set.  

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.