Low-volatility investing has emerged in the wake of demand for equity-like returns without the tail-risk of traditional equity. Low-volatility investing was first identified in the early 1970s by Fischer Black and Myron Scholes, and later reaffirmed by Eugene Fama and Kenneth French in 1993.
Academic research has shown that over the long run one can achieve benchmark-like returns with lower volatility (about two-thirds to a quarter of the volatility).
This appears to fly in the face of modern portfolio theory, which holds that risk and return are positively correlated. This investment anomaly also holds true across asset classes and spans almost every major market including the US, UK, Japan, Australia, Germany and Canada.
The graph (click for full view) shows the annualised return of the 1,000 largest US stocks from December 1979 to September 2012 based on their volatility decile. It shows the low-volatility anomaly (that low-volatility stocks outperform more volatile counterparts).
This anomaly is largely due to the structure of the investment industry, in particular the widespread adoption of relative benchmarks, as well as behavioural investing factors that led investors to make seemingly irrational decisions.
Investing in practice
Various strategies have been designed to produce a portfolio that delivers less volatility than the broad equity market. The most common include strategies that are limited to a low-volatility universe, driven towards minimal variance using a predicted risk model, or those positioned to meet a beta or other volatility target.
These strategies produce portfolios that are heavily weighted towards the least volatile 20% of stocks that do not rank among the 20% least volatile stocks.
In our view, these are based on a narrow interpretation of the low-volatility anomaly, which asserts that the lowest quintile stocks based on volatility outperform the market over the long run. The 20% least volatile stocks might be equal-weighted or weighted based on each name’s level of volatility. The S&P 500 Low-Volatility Index is an example of such a low-volatility strategy.
Minimum variance strategies typically include a broader universe of stocks than the least volatile 20% and employ a risk optimiser, and incorporate basic constraints regarding industry, sector and country weights to produce a portfolio with a predicted minimum possible volatility.
MSCI offers several minimum variance indices that utilise this method. Simply put, a true minimum variance portfolio would include 100% of its weight in the least volatile stock of the eligible universe.
As constraints are added (e.g., a maximum individual weight of 5%), the optimiser adds more positions to the portfolio with a goal of minimising the portfolio’s volatility. Beta-target portfolios are not limited to the least volatile 20% of stocks, nor are they necessarily driven by a risk optimisation engine, but are rather designed to produce a beta that is much lower than a corresponding index.
These portfolios can have some high-volatility stocks, but are weighted towards lower-beta names. Although there is some correlation between beta and volatility, the idea behind beta-target portfolios is that they will follow the direction of an index more closely than a targeted-volatility approach.
Targeted volatility strategies utilise a predictive risk model to construct a portfolio that has an acceptable level of risk. This can be an absolute or relative level of volatility. These strategies can use a broad set of stocks, incorporate constraints for industries, countries and regions, and – depending on the risk optimiser’s capabilities – can constrain other biases such as size and style.
Regardless of whether a low-volatility method is based on a strict interpretation of the low-volatility anomaly, a minimum-variance approach, a beta target or a volatility target, these strategies tend to produce portfolios that share similar characteristics.
First, they share overweight positions in many of the same low-volatility names, as well as underweight positions in the same high-volatility names.
Second, even if the method is not explicitly beta-targeted, these portfolios tend to arrive at the same beta of around 0.60. Third, whether or not the method is volatility-targeted, these portfolios tend to reduce volatility over the long run by 25-35%.
Fourth, because certain industries tend to fall within the low-volatility universe, these portfolios typically share a dividend yield that is higher than the respective indices.
Simulated, as well as real-world, examples show that these features can be maintained over the long run while achieving a return similar to major indices.
These strategies also often share some less desirable features, including concentrated sector weights, concentrated positions, extreme biases in small-capitalisation stocks and large weights in stocks that have low liquidity, as well as weak analyst coverage.
Exhibit 2 (click for full view) illustrates a typical sector bias associated with low-volatility strategies. It shows that 55% of the weight in this portfolio was concentrated in utilities and consumer staples.
The lack of diversification inherent in the concentrations described above exposes the portfolio to unpredictable risk. We have seen, for instance, that industries and countries rotate from high-volatility to low-volatility and vice-versa.
Prior to 2007, banks were regarded as low-volatility. An unconstrained, low-volatility portfolio would have been dangerously exposed to the financial sector during the global financial crisis.
The Japanese tsunami example is also illustrative. Prior to this disaster in March 2011, Japan and the utilities sector there appeared attractive based on volatility, and a global low-volatility strategy would likely have been overweight Japan and utilities.
Low-volatility investing 2.0
Given concentration tendencies, we believe a low-volatility portfolio is likely to perform better over a longer time-horizon if it is broadly diversified and includes liquid stocks that are well-known by analysts, and also incorporates stock-level research, rather than being exclusively driven by a desire to invest in the least volatile stocks.
To achieve this, some suggest avoiding high-volatility stocks, employing traditional rules of portfolio diversification, incorporating fundamental input, and optimising similar to a minimum variance framework.
Avoiding high-volatility stocks
Our research has shown that high volatility tends to be concentrated in about 40% of stocks within a universe, a phenomenon that occurs fairly consistently within global regions. This same group of stocks tends to underperform over the long run.
For the remaining 60% of stocks, volatility is distributed more evenly over the long run. During market declines, however, the least volatile 20% outperform more dramatically and with significantly less volatility.
Rather than using an approach that is driven by a small number of stocks focused towards the lower-volatility quintile based on their strong down-market performance, we believe that eliminating the most volatile 40% of stocks allows an investor to benefit from a better risk-adjusted return over full market cycles.
By eliminating the high-volatility stocks, we can design a portfolio that takes advantage of the remaining 60% of the investment universe to construct a more robust low-volatility portfolio.
Exhibit 3, below, shows the change in market volatility driven by the most volatile 40% of stocks in a universe of 1,000 of the largest US stocks over two 15-year time periods.
Enforcing rules of diversification
To ensure that certain industries, sectors or regions do not dominate a portfolio, some suggest constraining portfolio exposures to long-term market averages rather than the most recent market capitalisation.
The result is a portfolio that maintains some exposure to all investable areas of the stock market while minimising exposure to potential near-term market bubbles.
Risk optimisers also provide the ability to control other biases – such as size, liquidity and style – that otherwise can produce a portfolio no longer consistent with the universe it represents. Some suggest including rules that ensure the portfolio is consistent with the index it is measured against.
A typical low-volatility strategy makes no assumptions about the relative attractiveness of each company. Nor does it necessarily distinguish between the liquidity of stocks or how well they are covered by analysts.
Instead, the focus is solely on a stock’s volatility without regard to an expected return forecast. We believe this is a shortcoming of these strategies.
Adding fundamental analysis to a low-volatility universe offers the opportunity to generate additional return for an investor. Not all low-volatility stocks will have the same prospects, risks, balance sheets or liquidity.
Given that small-cap, less liquid and weakly covered stocks can represent a large portion of a typical low-volatility portfolio, we believe that incorporating fundamental research is not only advantageous, but critical.
Optimising in a minimum-variance framework
Once the universe has been established, diversification constraints have been determined and information from fundamental research has been applied, one has the framework required to optimise a portfolio to a desired level of risk.
Risk optimisers typically employ a multi-factor risk model that considers sources of systematic risks – such as industry, capitalisation, valuation and financial leverage – as well as a company’s unique or idiosyncratic risk.
Depending on the investment universe and investment constraints used in an optimisation, minimum variance strategies achieve a level of volatility for US stocks that are about 25-30% lower than the S&P 500. In a global portfolio, volatility can be reduced even more due to the larger universe of stocks available for optimisation.
What it means for investors
The strategies discussed in this article may offer ample opportunity for superior risk-adjusted performance through stock selection, risk management and portfolio construction with a broader universe of less-volatile securities.
Due to higher tracking error (but lower absolute risk) and the episodic investment performance of low-volatility portfolios, investing in such strategies requires a long-term perspective that may allow one to weather the underperformance in bull markets balanced by the outperformance in bear markets.
The statements and opinions expressed in this article are those of the authors and do not necessarily reflect MFS’ current views, and no forecasts are guaranteed. The securities and/or sectors mentioned in this article should not be viewed as advice, a recommendation, a trade indication or advertisement to buy, sell or hold the same. The article is intended for institutional use only. Issued in Hong Kong by MFS International (Hong Kong) Limited (“MIL HK”), a private limited company licensed and regulated by the Hong Kong Securities and Futures Commission (the “SFC”). MIL HK is a wholly-owned, indirect subsidiary of Massachusetts Financial Services Company, a US based investment adviser and fund sponsor registered with the US Securities and Exchange Commission. MIL HK is approved to engage in dealing in securities and asset management regulated activities and may provide certain investment services to “professional investors” as defined in the Securities and Futures Ordinance (“SFO”). For investors in Australia: MFSI and MIL UK are exempt from the requirement to hold an Australian financial services license under the Corporations Act 2001 in respect of the financial services they provide. In Australia and New Zealand: MFSI is regulated by the US Securities & Exchange Commission under US laws and MIL UK is regulated by the UK Financial Conduct Authority under UK laws, which differ from Australian and New Zealand laws.