On Liabilities in Asset Allocation

From an e-mail from one of my readers:

I’m not sure if you have the time to respond to this, but figured I’d send to you and just see!…

(Just FYI, I’m not an investment professional of any sort, so I don’t have any “skin in the game” as they say, just a geek who follows the markets and DIY financial-planning issues and long-time reader of the Aleph blog)

I recently read an FP article by a guy I’ve read a lot (Alan Roth).  He suggested that, when your analyzing an investment portfolio and making asset-allocation decisions, you need to treat mortgage debt as an “inverse-bond” or an “anti-bond”…such that any mortgage debt held would dollar-for-dollar negate or reduce your actual bond investment holdingsAnd the result is that this made the investor’s actual portfolio risker than they realized, since their “true” bond allocation was smaller than they had considered.

I thought it was a novel concept, but I found some problems with that approach, within the context of asset-allocation.  My main point was that the primary purpose of analyzing a portfolio’s asset allocation is to manage risk through diversification of assets (generalizing here in interests of being concise).   The pinnacle of diversification is non-correlation: generally in economic environments where equities soar, bonds will underperform, and vice versa.  However, classifying a debt as an “anti-bond” doesn’t actually provide any portfolio diversification, or introduce any non-correlation.  It won’t actually negate the amount that your bonds would rise, and it won’t actually offset the amount your bonds would fall, in those respective market environments.  And even if you consider that the real value of the debt is decreased if inflation rises (as the NPV calculation would be using a greater discount rate), that doesn’t have any real-world effect on the portfolio and it’s risk and return behavior.  Since borrowers aren’t allowed to “mark-to-market” their mortgages, that debt holding value does not fluctuate–it is fixed, and amortized from its historical cost, regardless of any market conditions or any theoretical NPV/DCF changes.  Therefore, the inverse- or anti- bond holding in the portfolio has zero impact on the portfolio’s actual risk/return behavior, and so it seems to me it doesn’t add any functional value to frame debt as an “anti” portfolio holding of some sort.

Also, if you were going to do that, to be fair and complete, you must apply that same principle to every single debt the client has (otherwise, it would be rather arbitrary just selecting the mortgage debt).  This adds unnecessary complexity in the asset allocation analysis.

Instead, the appropriate (and only) way to analyze debt is, separate from investable portfolio assets, on the cash-flow side of things.  Simply asking what is the “optimal” use of the available capital; i.e. what net “return” do you earn by using capital to eliminate debt, versus what net return could you earn if you kept the debt and employed the capital elsewhere (this will be different for each investor and their unique situations).  This is the way to analyze and evaluate debt, not to mingle it in with your invested assets and classify it as an “anti-bond” holding within your portfolio.

I was just curious your take on this, and if I am misunderstanding or missing something.  Do you ever consider client’s debt as “inverse-“ or “anti-“ bonds in the context of asset allocation?


When you manage money financial firms, if you do it right, you consider the promises that your firm needs to fulfill.  When will cash be needed to pay obligations?  That helps drive asset allocation, because assets should broadly match liabilities.

Now, I am not a financial planner.  That said, the same principles apply to personal asset allocation.  If someone has a large mortgage or other debts, and he can’t invest his fixed income assets at levels that exceed the yields on those debts with reasonable risk, he should not invest in bonds — he should pay down his debt.  In the case of 401(k)s or IRAs, where there might be matches or tax advantages, the calculation becomes more complicated.  You have to weigh the match and tax deferral vs the negative arb on bond yields vs the mortgage and personal debts.

There is another factor here — how stable is your job?  If stable, it is bond-like, and you can take more equity risk with investments.  If your job has payoffs that vary a great deal with the market — commissions, bonuses, etc., it is stock-like, and you should take less risk in your investing — take excess earnings and pay down the mortgage.  I did that when I went from being a bond manager inside an insurance company, to being an equity analyst inside a hedge fund.  I paid off my mortgage in full, so that I would be free to take risks for my new employer.

As for the article, the concept is not novel.  It is well-known and practiced by institutional asset managers who manage money to the horizons needed by their clients.  As an example, the cash flows of a pension plan are relatively determinate, and the discount rate calculates the value of the liability.  The portfolio should throw off cash when needed in order to minimize risk.

In some cases, where bonds don’t offer enough yield, and equity prices are depressed, it might make sense to tactically mismatch, betting that equities will offer better returns versus the liabilities than bonds would on a risk-adjusted basis.

This argument has made its rounds for the last 20 years in insurance and pension management?  Do we match asset and liability cash flows, or do we trust in the equity premium, and invest in risk assets?  The correct answer is hybrid.  In general, match assets and liabilities, but if there is a significant tactical advantage to not match, then do that.  Think of buying junk bonds in late 1998.  Time to throw matching out the window.  And then in mid-1999, buy equities.

Now, not all clients will allow for that much risk-taking.  Many institutional investors will not let the asset manager take advantage of temporary dislocations.

In general, I think Mr. Roth is correct, but with an adjustment.

  • In extraordinary times, where bonds yield more than the earnings yields of stocks (think 1987 & 2000), buy bonds heavily, even if you have mortgage and other debts.
  • In extraordinary times, where stocks earnings yields are much higher than bonds, mismatch and own more stocks relative to bonds.  Just beware deflation, with falling future earnings.
  • In normal times, an indebted investor should not add to his leverage, but should invest in bonds, or better, pay down his debt.  Being debt-free is an excellent thing, and allows the investor to take more risks when the market is offering bargains.

Debt is either something to be funded by bond assets, or funds a margin account where you outperform the yield, or die.  All of this depends on where the market is in its risk cycle.  Only take risk where it is rewarded.