What is the “Low Volatility” Factor?
A long-standing theory of financial markets is that higher risk demands higher reward.
The evidence behind the low volatility anomaly stands in direct opposition to this theory. At its core, the anomaly (sometimes called the closely-related “bet-against-beta,” or “BAB,” factor) captures the empirical evidence that low risk investments have higher returns and high risk investments have lower returns than would be expected from standard financial theory.
While initially identified as an equity effect, research has found the anomaly in many areas:
· Haugen and Heins (1972) provided empirical evidence that from 1926 – 1969, portfolios formed on low-volatility U.S. equities exceeded the return expected by their low level of beta
· De Carvalho, Dugnolle, Lu, and Moulin (2014) suggest that the factor can be applied to major fixed income markets and regions
· Frazzini and Pedersen (2014) find the effect works when applied across assets in commodities, currencies, equities, and fixed income
Popular Equity Implementations
Today, investors looking to tap into the anomaly in the equity space have a number of options. Consider three of the most popular low volatility strategy ETFs:
- The PowerShares S&P 500 Low Volatility ETF (SPLV)
- The SPDR SSGA US Large Cap Low Volatility ETF (LGLV)
- The iShares Edge MSCI Min Vol USA ETF (USMV)
All three take different approaches to portfolio construction: differences that investors should be aware of before making their investment decision.
SPLV: Bottom-up & Unconstrained
For SPLV, the general outline of the methodology is (source: methodology document),
- Rank stocks within the S&P 500 by their 12-month realized volatility
- The 100 stocks with the lowest realized volatility are selected
- Weight each stocks in proportion to its inverse realized 1-year volatility
Simple, transparent, and most in line with original academic research into the factor, this is an implementation of a low volatility portfolio at the index level.
This construction can lead to significant sector and industry bets, which some argue confounds whether the success of such an approach is actually due to the low volatility effect or merely unintentional sector bets that have paid off historically.
Fortunately, Asness, Frazzini, and Pedersen (2014) find that when applied across industry groups, the low volatility effect persists. So even if the approach ultimately leads to the selection of the lowest volatility industry groups, there is empirical evidence that such an approach has merit.
LGLV: Bottom-up & Sector-Constrained
For LGLV, the general outline of the methodology is (source: methodology document),
- Assign stocks from the Russell 1000 to their respective 10 sectors
- Rank stocks within their sector based on trailing 60-month realized volatility
- For each sector, pick the lowest volatility stocks until total free float market cap reaches 30% of the sector total
- Weight each stock in proportion to its inverse realized 60-month variance (i.e., 1 / volatility squared), constrained to the lesser of 5% or 20x the index weight.
This results in a portfolio that has sector weights that align with the Russell 1000, but the underlying securities within each sector differ dramatically.
The important distinction is, based upon the definition above, this is an implementation of the low volatility factor at the sector level, not the index level.
Does such an approach work?
Asness, Frazzini, and Pedersen (2014) find evidence to support the approach while De Carvalho, Zakaria, Lu and Moulin (2014) find that sector-neutral, low-volatility approaches may actually be more efficient at harvesting alpha than non-sector neutral approaches.
USMV: Top-Down & Constrained
For USMV, the methodology is less transparent (source: methodology document) as the approach is optimization based.
The general idea is re-weight the holdings within the MSCI USA index, subject to a number of constraints on both holding and sector concentrations, to create the minimum volatility portfolio. The constraints, however, are loose: sectors need only be within +/- 5% of the index level and securities can be up to 20x their initial weight (though capped at 1.5%).
This approach is unique in that it is top-down: securities are selected in concert with one another to create a portfolio-level effect. The question is, does a portfolio with minimum volatility necessarily imply a tilt to a low volatility effect? Is it not possible to construct a minimum volatility portfolio by combining high risk, but negatively correlated, securities?
As it turns out, when you are considering a non-trivial example, the answer is largely “no.” In fact, De Carvalho, Lu, and Moulin (2012) find that the key factor exposures in a minimum variance portfolio is low beta and low residual volatility stocks. So despite a more opaque construction methodology, USMV should tap into the same low volatility factor as USMV and LGLV.
Conclusion
Which approach reigns supreme?
It depends on your views relating to sector and security concentration. Each approach introduces its own risks:
- SPLV’s approach can lead to large sector concentration bets
- LGLV, though constrained from a sector perspective, may introduce large security bets through its inverse-variance weighting scheme
- USMV sits between the two, with greater sector constraints than SPLV but less than LGLV, and potentially more concentrated holdings than SPLV but less than LGLV
The question of which approach to employ will largely rest upon which risks investors feel comfortable taking.
The good news is that no matter which approach they take, all three construction methods have a solid grounding in academic evidence that supports their exposure to the low volatility anomaly.
Disclosure: Newfound currently utilizes USMV within its U.S. Factor Defensive Equity strategy.
References
Asness, C.S., A. Frazzini, and L.H. Pedersen. “Low-risk investing without industry bets.” Financial Analysts Journal, Vol. 70, No. 4 (2014), pp. 9-12.
De Carvalho, R.L., X. Lu, and P. Moulin. “Demystifying Equity Risk-Based Strategies: A Simple Alpha plus Beta Description.” The Journal of Portfolio Management, Vol. 38, No. 3 (2012), pp. 56- 70.
De Carvalho, R.L., M. Zakaria, X. Lu, and P. Moulin. “Low-risk anomaly everywhere: Evidence from equity sectors.” SSRN No. 2527852, 2014.
Frazzini, A., and L.H. Pedersen. “Betting against Beta.” Journal of Financial Economics, Vol. 111, No. 1 (2014), pp. 1-25.
Haugen, R.A., and A.J. Heins. “On the Evidence Supporting the Existence of Risk Premiums in the Capital Markets.” Working paper, SSRN No. 1783797, 1972.