If you could shrink the size of your bond allocation, but maintain the same dollar and duration-weighted allocation, and free up some of that capital to put to use in other asset classes as a result, how would that sound? I wanted to present something that might be controversial in the field of portfolio allocation.
I believe that a form of bond leverage is available in the form of duration. By holding a longer average duration of bonds, the bond portion of a portfolio can then be smaller. In effect this is free leverage. Free in terms of cost, but does hold some risk. Longer duration bonds are more volatile and can expose the investor to a larger risk of loss. However, this is accounted for by having a smaller overall bond position. Additionally, the conditions that may make bonds go down would also be conducive to a favorable stock market, in terms of occurring in conditions of economic growth.
A model 60/40 ETF portfolio
A simplified baseline portfolio is a 60/40 portfolio, which is to say 60% equities and 40% bonds for simplification. (I won't go into international vs U.S. equities exposure or alternative assets, in this simplified analysis). Using an ETF approach, one could have 60% in the Vanguard total US stock market ETF (NYSEARCA:VTI) and 40% in the Vanguard total bond ETF (NYSEARCA:BND). BND has an average portfolio bond duration of 5.2 years (Current as of 01/09/2013).
Exploring a "leveraged" long-duration bond portfolio
With one simple change, of substituting the Vanguard Extended Duration bond ETF (NYSEARCA:EDV), which has an average duration of 26.2 years, one can obtain leverage. Note that I don't use the word leverage to imply a recommendation to go "all-in" on bonds. Quite the opposite: I am talking about using leverage to enable taking SMALLER dollar-size position in bonds. This will become more obvious after we look at the relative volatility (beta).
In looking at the positive and negative volatility of comparing EDV to BND on a one-year price chart, the longer duration of EDV is about five times that of BND, which is approximately proportional to its five times longer bond duration. To be more quantitative about it, I found that Morningstar reports a beta of 6.6 for EDV, versus 1.05 for the more average duration BND. Again, that's approximately five-six times the beta, which mirrors my back-of-the-envelope chart analysis, and also happens to approximate the relative difference in portfolio durations of each fund.
Comparing the correlation coefficient of the US stock ETF, VTI, to the two bond ETFs, BND and EDV reveals an even more negative correlation of EDV to the stock fund of -0.81 versus -0.67. Keep in mind that this really just captures the directional movement and not the magnitude, so this analysis cannot be used to determine position sizing.
Putting it all together
So in theory, based on the relative duration, and my eyeballing of the stock charts, and the more quantitative formal evaluation of relative beta, the bond portion could be reduced in amount if substituting a longer-duration ETF like EDV. My first guess hypothesis, before evaluating the volatility and betas, was that it could be reduced from 40% bond to 20% bond. However, quantitatively, based on the beta difference, one could argue about taking the bond component as low as 6% allocation to EDV (40% bond allocation / 6.6 beta = 6%) and yet maintain the same bond allocation. Obviously that assumes the beta will be maintained in future market conditions, and of course there is no way of predicting that.
This approach would free up 34% of the total portfolio to re-allocate toward other asset classes, while maintaining the same bond exposure. That, my friends, is a form of leverage. And it doesn't cost anything - not even additional risk (in theory) because you would be taking down the dollar size by 34% to account for the known higher volatility and beta of the long-duration bond ETF. Theories can turn out to be wrong at inopportune times, but it is hard to see how risky a 6% allocation would be - even if it went to zero that's still only 6% of your portfolio. If long-term bonds could even go to zero, we'd be living in a pretty ugly world.
One could either re-allocate all of the remaining 34% into the equity portion, or all into an alternative investment (such as commodity ETF, REITs, MLPs) or a money market fund, or some combination of these. I am not sure I like the idea of reallocating the balance all to equities, since it would increase risk from that asset class.
On the other hand, I like the idea of allocating some into cash and some into alternatives, as this provides some additional diversification for the same-size portfolio, while retaining all of the original bond exposure. Note that some of the examples I listed above have their own interest-rate sensitivity so you'd need to consider how that could interact with a long-term bond ETF to make sure they could peacefully co-exist in your portfolio.
This report is intended to show a way in which you can reduce the dollar size allocation to bonds by substituting a bond fund that is more volatile due to its longer duration. As noted earlier, many people are appropriately concerned about the risks of bonds, since they will lose value if we experience high inflation or an economic growth rebound. Despite that risk, which is always present, most experts recommend continued exposure to bonds as a way of diversifying and dampening portfolio volatility. I am merely trying to show a way to do that with a smaller bond position size, which would allow allocation to yet other asset classes and provide additional diversification, without employing overt leverage or margin.
I hope you found this useful for generating ideas, and I look forward to the discussion of any thoughts you might have.
Disclosure: I have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.