At one point, low volatility and minimum variance ETFs were highly popular and supported by academic research that identified an anomaly in which low-risk stocks produced higher returns than high-risk stocks.
This finding runs contrary to Modern Portfolio Theory and the Capital Asset Pricing Model. One of the seminal papers identifying this anomaly was Risk and the Rate of Return on Financial Assets: Some Old Wine in New Bottles (Haugen and Heins, 1975).
A follow-up working paper written by Robert Haugen and Nardin Baker in 2012 included a comprehensive analysis of 21 developed markets and 12 emerging markets from 1990 to 2011 showing that stocks in the lowest volatility decile outperformed those in the highest volatility decile on a total return and risk-adjusted basis.
This paper was no doubt welcomed by ETF issuers looking to launch and market low volatility-based ETFs.
Now it takes a ‘volatile’ situation for investors to become interested in low volatility strategies. Indeed, it was not until 2011, after the Global Financial Crisis, when ETF issuers began launching low volatility-based ETFs.
As for index providers, MSCI launched their first minimum volatility index in 2008 and S&P launched the S&P 500 Low Volatility index in 2011 and the US-listed Invesco ETF tracking this index now has a market value of $7.5bn.
Both indices have different methodologies but what they have in common is that they’ve underperformed the S&P 500 since their respective launch dates.
This is not an uncommon finding because relationships between factors including volatility, size, value, quality and momentum have a time-varying component resulting in relative performance being cyclical.
Certainly, low volatility ETFs can be a useful part of a dynamic asset allocation process, but it is helpful to understand the differences in their underlying methodologies and characteristics before investing and that is what the rest of this article sets out to do.
UCITS ETFs in our analysis
The ETFs included are listed in Chart 1 and were selected based on length of track record and to highlight different underlying index methodologies and fundamental variations (ESG, high dividend etc.).
Different methodologies and active risk
Three main underlying index methodologies are employed across these six ETFs. The weighting methodologies have a big impact on the active sector exposures and ultimately tracking error relative to a broadly diversified, market cap weighted benchmark like the S&P 500 (see Chart 2).
Low volatility indices use an inverse volatility weighting whereby stocks are ranked based on their trailing volatility and those with a lower (higher) volatility receive a proportionally higher (lower) weighting. For example, the S&P 500 low volatility index only includes the 100 lowest-volatility stocks. You are not getting a low volatility version of the S&P 500 but rather an exposure to the 100 least volatile stocks selected from the S&P 500. The MSCI low volatility indices take a more stratified approach to dampen the level of tracking error relative to the parent index.
Minimum volatility indices use a minimum-variance optimisation to determine the index composition and weights. Constraints are imposed at the security and sector level to restrict the active level of risk versus a broad, market cap benchmark.
Minimum volatility with additional risk factor constraints. In addition to minimising variance, these indices include additional risk factor constraints as part of the optimisation objective function to control for market beta, size and style tilts.
The result is a minimum variance ‘lite’ exposure with less active risk relative to the parent index.
Rolling three-year annualised tracking error relative to the S&P 500
Tracking error has increased for all the ETFs as sector dispersion increased in recent years. The lowest tracking error ETFs use minimum-variance optimisation weighting with sector, security and factor constraints.
The highest tracking error ETFs use inverse volatility weighting and in the case of the Invesco ETF, also have a high dividend selection criterion (see Chart 3).
Rolling three-year annualised Sharpe ratio
One of the characteristics of lower volatility indices is that they tend to deliver higher Sharpe ratios than the market index.
For example, over rolling three-year periods going back to 1991, the MSCI USA Minimum Volatility index and S&P 500 Low Volatility index had higher Sharpe ratios than the S&P 500 over 71% and 54% of rolling three-year periods, respectively.
However, over more recent periods, this has not been the case for ETFs tracking these or similar indices as illustrated in Chart 4. Long-term risk factor exposures of each low volatility focused ETF A common characteristic of low volatility ETFs is a relatively low equity market beta.
In Chart 5, the equity market betas range between 0.68 to 0.84 versus an S&P 500 market beta of 1.0.
Additionally, all the ETFs have positive exposure to the value premium and unsurprisingly, the Invesco S&P 500 high dividend low volatility has the highest exposure to the value premium.
This ETF is the only ETF with a positive exposure to the small-cap premium whereas the other ETFs have a negative exposure which means they have a large-cap bias.
Lastly, the ETFs have very little momentum exposure. Collectively, this information is useful in understanding how these ETFs may fit into a portfolio and align with an investor’s risk profile and views on the market.
Low volatility factor ETFs returns are sensitive to changes in business cycle variables
When evaluating the risk and return behaviour of these six low volatility-based UCITS ETFs, Parala’s model captures not only each ETF’s exposure to four important risk factors including market beta, small cap, value and momentum risk premiums but also the fact that the performance of the risk factors themselves are sensitive to the business cycle.
Each ETF also has an important residual component of its return which is independent to the risk factors but is correlated with the business cycle and may be due to differences in each ETF’s sector weights and security selection and weights within each sector.
We can identify these relationships using important macroeconomic variables like the default spread, term spread, shortterm interest rate, dividend yield, VIX index and commodity prices.
Based on their sensitivities to these macroeconomic variables we gain additional insights into their likely behaviour across the business cycle. For example, all the ETFs have a positive sensitivity to the default spread so if credit risk increases, these ETFs would be expected to benefit.
Notice that the Invesco S&P 500 High Dividend Low Volatility ETF has a positive sensitivity to the term spread whereas the iShares Edge S&P 500 Minimum Volatility ETF has a negative sensitivity.
In an environment where the term spread is inverting as we have experienced recently, the iShares Edge S&P 500 Minimum Volatility ETF would be expected to perform better and that is indeed the case.
The same would be expected in a rising rate environment which would benefit the iShares Edge S&P 500 Minimum Volatility ETF and weigh on the S&P 500 High Dividend Low Volatility ETF.
Overall, the current macro environment would appear to favour the iShares Edge MSCI USA Minimum Volatility ESG ETF and been least favourable for the Invesco S&P 500 High Dividend Low Volatility ETF and their relative performance seems to confirm this (see Chart 6).
Conclusion
It takes a ‘volatile’ situation for low volatility-focused ETFs to show their merits which is why these ETFs may have lost some of their short-term appeal. In low volatility, strong economic environments, growth investing reigns supreme.
However, low volatility-focused ETFs may be useful to adjust sector, beta, volatility and macroeconomic diversification within a portfolio and for that reason represent a useful tool in the toolbox.
These low volatility UCITS ETFs employ different weighting methodologies, have different active sector, risk factor and macroeconomic exposures which can inform investors how they might behave in different market and macro environments.
Having a sophisticated model like Parala’s may be helpful when making investment decisions but simply understanding each ETF’s characteristics, risk factor exposures and business cycle sensitivities can be useful to help professional investors select funds and build portfolios that align with their ‘house’ views.
Steven Goldin is managing partner at Parala Capital, which provides institutional investors and asset owners with asset allocation advice using advanced quantitative technology based on the academic research of its founding partners
This article first appeared in ETF Insider, ETF Stream's monthly ETF magazine for professional investors in Europe. To read the full edition, click here.
