The principle behind a traditional 60/40 investment portfolio is balancing two asset classes - large-cap U.S. stocks and U.S. bonds. The rationale is that U.S. stocks and bonds have low performance correlation - historically, bond funds seldom have had negative annual returns, while stock funds lost money in a calendar year nearly 30% of the time.
All bond funds, of course, are not created equal in terms of their raw performance or correlation to the stock fund with which they're teamed up in a 60/40 portfolio. Investors understandably may seek a bond fund with the best performance. But if the performance of the "best" bond fund is highly correlated with the performance of the stock fund it's paired with, the logic of a 60/40 portfolio is thrown out of whack. The goal is to add an asset class that behaves differently than stocks to provide a counter-balance when stocks tank, not to find a bond fund that produces returns that mimic a stock fund.
Investors would be well served to emphasize the rebalancing premium - the extra return generated when a 60/40 portfolio is regularly rebalanced. It turns out that the size of a rebalancing premium is related directly to the correlation between the stock fund and bond fund, which in turn predicts whether a particular bond fund is the right fit for a balanced portfolio.
A DEEPER LOOK
Think about how a typical 60/40 portfolio operates by considering the performance history of large-cap stocks (S&P 500) and bonds (Ibbotson U.S. Intermediate-Term Government Bond Index) over the 85-year period from 1926 to 2010 (see "Yin and Yang?" on page 130). Bonds had a negative annual return on eight occasions, about 9% of the time. The losses were relatively small, never exceeding 3%. Over the same period, stocks were in the red 24 times, or 28% of the time. Six of the annual losses exceeded 20%.
The motivation to hold stocks in a balanced portfolio is based on the higher expected return. Indeed, the average annualized return for large-cap U.S. stocks over the 85-year period was 9.9%, versus 5.5% for U.S. bonds. During the period, the worst three-year slide for the S&P 500 was a 61% free fall, but for bonds the worst cumulative return over any three-year period was 1.6%, illustrating the pragmatic difference between stocks and bonds.
There is another important element to a balanced portfolio: low correlation between return patterns. The key to harvesting the benefits of low correlation between assets in a portfolio is systematic rebalancing to the 60-40 ratio. This process involves a flow of funds between the two kinds of assets, meaning rebalancing is most profitably done in tax-sheltered accounts. Alternatively, cash inflows can be used to accomplish rebalancing.
In reviewing the behavior of the U.S. equity markets and U.S. bonds over the 10 years ending Dec. 31, 2010, the 10-year average annualized return of the Vanguard Total Stock Market Index (VTSMX) was 2.5%, which reflects a U.S. portfolio of 70% large-caps, 20% mid-caps and 10% small-caps. By comparison, the 10-year annualized return of the Vanguard Total Bond Market Index (VBMFX) was 5.6%.
The correlation of annual returns between these indexes was -0.37. Recall that the range of correlation coefficients is -1.0 to +1.0, where -1.0 represents perfect negative correlation and +1.0 indicates perfect positive correlation. The correlation of the monthly returns of these two funds was -0.09 over the 10-year period. One of the goals of asset allocation is to build a portfolio with low correlation, such as between -0.5 to +0.5. A reading of -0.37 indicates low correlation.
A 60/40 portfolio comprised of 60% VTSMX and 40% VBMFX achieved a 10-year annualized return of 3.7%, assuming each position wasn't rebalanced over the 10 years. If, however, this two-asset portfolio was rebalanced to 60/40 allocations at the end of each year, the 10-year annualized return was 4.35% - demonstrating the rebalancing premium. In this case, it was 65 basis points. It turns out that the size of a rebalancing premium is related directly to the correlation between the two assets; not all bond funds are created equal.
Bond funds with higher correlation to stock funds (in this case, VTSMX) tend to have a smaller rebalancing premium, whereas bond funds with lower correlation to the stock fund they are paired with have a higher premium. That means when combining a stock fund and bond fund in a portfolio that you intend to rebalance, you'll get better risk-adjusted performance by selecting a bond fund that's demonstrated lower correlation to the stock fund.
THE BOND FUND MATTERS
The chart in "The Rebalancing Premium" (see left) shows evidence of this low correlation rebalancing premium. All taxable intermediate U.S. bond funds with at least 10 years of performance as of Dec. 31, 2010, were extracted from Morningstar Principia (only distinct funds were selected). A total of 231 funds met the criteria. The average 10-year annualized return of these funds was 5.6% - the same return as VBMFX. The highest 10-year return among the group was 9%, while the lowest was 1.2%. The average 10-year correlation between this group of bond funds and VTSMX was 0.24%. The highest correlation was 0.86 (Invesco Income A) and the lowest correlation was -0.88 (Old Westbury Fixed Income).
Each of the bond funds was separately teamed with VTSMX in a 60/40 portfolio. Performance was measured under two different assumptions - no rebalancing over the 10-year period and annual rebalancing - and then the performance differential was compared. Because this group of bond funds had varying levels of return over the 10-year period, the analysis factored out the raw performance differences. The performance differential represents the rebalancing premium, which was plotted against the 10-year correlation between each fund and VTSMX.
The relationship between the bond fund and stock fund correlation and a rebalancing premium is extremely strong. There are 231 dots in the graph, each one representing a 60/40 portfolio in which all of the bond funds are teamed with VTSMX.
The dots in the upper left-hand corner have a higher rebalancing premium because those bond funds have lower correlation with VTSMX. The dots in the bottom right-hand corner had higher correlation with VTSMX and, therefore, a lower rebalancing premium. The rebalancing premium for the low correlation tandem of Old Westbury Fixed Income and VTSMX was nearly 80 basis points, while the premium for the high correlation pair (Invesco Income A and VTSMX) was 31 basis points.
"Bond Titans" (see above) highlights the 10 largest intermediate U.S. bond funds. Each fund was teamed with VTSMX in a 60/40 allocation and evaluated under the same two scenarios - no rebalancing and annual rebalancing.
For example, Pimco Total Return had a 10-year annualized return of 7.34% through Dec. 31, 2010, and had a 0.10 correlation with VTSMX. If that 60/40 tandem was never rebalanced, the annualized return was 4.4%. However, if they were annually rebalanced, the 10-year annualized return was 5%, meaning Pimco Total Return demonstrated a 61 basis point rebalancing premium. Of the 10 biggest bond funds, American Funds' Bond Fund of America had the lowest rebalancing premium at just 36 basis points - and one of the highest correlations to VTSMX at 0.79.
Here's a practical application: You're building a 60/40 portfolio you intend to rebalance regularly. When choosing between two bond funds with similar returns to team with a stock fund, choose the bond fund with lower correlation with the stock fund you're selecting. By doing so, you increase the likelihood of a rebalancing premium.
Craig L. Israelsen, PhD, is an associate professor at Brigham Young University. He is the designer of the 7Twelve Portfolio (7TwelvePortfolio.com) and author of 7Twelve: A Diversified Investment Portfolio with a Plan.