As an active participant in designing portfolios, I try to be cognizant of best practices and what portfolio design techniques will predictably deliver desired results. Therefore, I was pulled into a recent string of articles initiated by fellow SA contributor, Daniel Moser.
Mr. Moser has presented the idea that risk allocation based on volatility is a worthwhile endeavor and appropriate versus dollar allocation. (A follow up was provided based on commentary/criticisms by fellow contributor Roger Nusbaum.) I would agree largely with Daniel's perspective and would like to expand the notion for those willing to accept that volatility is not just for armchair academicians.
To be clear -- I view volatility as a real risk that should be accounted for in portfolio design both between asset classes, as well as within asset classes.
First, let me state that without reinventing the wheel -- Mr. Moser's thrust is a cornerstone of the concept of Risk Parity. Portfolio designs using Risk Parity are based on balancing the volatility of each asset class. This can be done by lowering the volatility of highly-volatile asset classes, or using leverage to increase the volatility of lower-volatile asset classes.
Volatility As A Potential
Volatility is really a potential risk of an asset class. Actual portfolio performance will be dictated by current and future circumstances. Volatility just defines the range of potential outcomes based on past experience. A simulation of potential outcomes can be created using the estimated volatility of the portfolio (along with other inputs). The higher the volatility, the greater the range of outcomes, which for downside risk analysis, increases the probability that a portfolio will not meet expectations.
The risk measurement tool, Maximum Drawdown (MaxDD), does a good job of bridging the gap between the volatility risk and identifying worst case scenarios based on past experiences. MaxDD is calculated as the largest decline (peak-to-trough) from an asset class' previous peak.
MaxDD will be greater for those asset classes with higher volatility. The S&P 500 (SPY) has a 3-year volatility measure of 15.6% compared to the Barclay's Aggregate Bond Index (AGG) of 2.73%. Over the past 10 years, the S&P500 MaxDD was ~55% on a total return basis. This compares to a MaxDD of only ~5.5% for the Barclay's Aggregate Bond Index.
Using Vanguard's Balanced Fund (VBINX) as a real-time proxy for the 60/40 Balanced Portfolio (60% Stocks, 40% Bonds), the MaxDD for the balanced fund was 35% over the same 10-year period. The 3-year volatility of the balanced fund is 9.36%. It makes sense that if 90% (95% argued by Mr. Moser) of the volatility is based on the SPY, then 90% of the maximum drawdown will be related to SPY. This happens to be the case.
[The following three figures show the MaxDD calculation for the S&P 500, a proxy for the Barclay's Aggregate Bond Index, and a 60/40 Balanced Portfolio. The blue line represents the percentage drop from the previous peak. When the blue line is at zero, the investment is at a new peak. I kept all three figures scaled to the SPY to better reflect the magnitude of the drawdowns.
(click images to enlarge)
Figure 1: S&P500 10-Year MaxDD
Data Source: Yahoo!Finance
Figure 2: Aggregate Bond Index (VBMFX) 10-Year MaxDD
Data Source: Yahoo!Finance
Figure 3: 60/40 Balanced Portfolio (VBINX) 10-Year MaxDD
Data Source: Yahoo!Finance
Volatility As A Factor (to be short)
As shown above, volatility is a large determinant in the potential magnitude of a decline in a portfolio. While Mr. Moser highlighted increasing the volatility of the fixed income portion, a portfolio's performance can also be improved by factoring out volatility within an asset class. Stand & Poor's has created a low volatility index based on the S&P 500, which is effectively short volatility. The lowest 100 volatile stocks of the S&P 500 based on the previous year are included in the index. (The index is rebalanced quarterly.) The following figure highlights index performance. (I would include real-time results, but the currently investable ETF -- the PowerShares S&P500 Low Volatility ETF (SPLV) -- does not have a long enough time period.)
Figure 4: Low-Volatility Index Comparison to S&P 500
Low Volatility (Ranking Based)
Annualized Standard Deviation
Source: S&P Dow Jones Indices LLC as of 3/31/12
Consider replacing the S&P500 SPDR with the low volatility index based ETF in a 60/40 Balanced Portfolio. The volatility of the portfolio would be reduced by ~30% based on back of the envelope math. Therefore, a portfolio's volatility can be balanced using higher-volatile fixed income investments (or using explicit leverage with a closed-end fund such as Alliance Bernstein's Income Fund (ACG)), or through lowering the volatility of the equity asset class.
Volatility is a worthwhile consideration for portfolio design, and not only for armchair theorists. MaxDD highlights the downside risk to the potential risk of volatility. The higher the volatile nature of the asset class -- the greater the MaxDD and the longer the time for the asset class to regain its previous peak.
Furthermore, volatility can also be actively factored within asset classes to improve the overall characteristics of the portfolio. Research shows that factoring an equity portfolio for low-volatility improves returns while lowering the volatility of the portfolio.
The conclusion, in my mind, volatility between asset classes and within asset classes are sources of risk that are worth actively managing for.