Friday, February 6, 2015

Long Ladders

Long TIPS bond ladders demonstrate the challenge of matching liabilities in the more distant future, say, funding the last half of a 30-year retirement.

This is one of the toughest pieces of retirement funding to figure out for several reasons. First, we don't know if we will still be alive when it begins, which is a major reason people don't like annuities. Retirees who don't live beyond their life expectancy won't get much benefit from an annuity.

Or, we might live that 15 years and then some, possibly outliving a bond ladder and wishing we had purchased the annuity. Inflation has a much greater impact on more distant years of spending, and even when inflation protection can be purchased it is quite expensive.

In short, the further into the future we plan, the more uncertainty we must deal with. Life annuities remove the uncertainty of living a very long time (longevity risk). Long TIPS bond ladders provide an alternative, but come with a different set of risks, including some degree of longevity risk.

I'm a big advocate of floor-and-upside strategies that secure an acceptable level of income before investing in a risky portfolio. A lot of really bright people, like Zvi Bodie (Risk Less and Prosper), Nassim Taleb (The Black Swan), William Bernstein (too many to list) and Wade Pfau (How Do I Build a TIPS Bond Ladder for Retirement Income?) like TIPS bonds in the safe "floor" portfolio.

TIPS bonds held to maturity are considered risk-free assets – they have no default risk, no interest rate risk, no inflation risk and no correlation to market returns – but no asset is absolutely risk-free. With a TIPS bond ladder, there is the aforementioned risk that you might live longer than the ladder you buy and there is also a risk that you won't be able to hold all of your TIPS bonds to maturity, despite your intentions, in which case you will have interest rate risk.

The risk that you will not be able to hold all the bonds to maturity is obviously greater for a 30-year ladder than for a 5-year ladder and that is one reason I separate this discussion of long ladders from the previous post addressing short ladders. Short ladders are about as risk-free as investments assets can be, but risk grows with the length of the ladder.

Here is the scenario that makes me waiver just a bit. Let's say I buy a 30-year TIPS bond ladder today. Since yields are at record lows currently, I will likely lock in low interest rates for the next thirty years and when interest rates begin to rise in a few years, which seems more likely than not, I will regret not having waited.

I shouldn't regret the purchase because I will ultimately get what I want, a near-certain match of those future liabilities. It's just that the price for this income will decline in this scenario and I'll feel like the guy whose neighbor gets a better deal on a car identical to his. It shouldn't make me feel any worse about my own car, but it does.

If I buy a TIPS bond ladder, my goal isn't to invest to optimize my return or to get the best deal (risk-free assets never achieve that over time), it's to provide certainty of future income.

A second concern I have is that I would need to buy several long bonds.

I hate long bonds.

They're almost as risky as stocks and their return doesn't adequately compensate for that risk. As I mentioned in my last post, Funds and Ladders: What Matters?, in 2013, a bad year for bonds, iShares intermediate ETF TIP lost 8.65%, while long duration (27) bond ETF PIMCO ZROZ lost 22% of its value.

Long bonds fall much faster in value when yields increase than short or intermediate bonds do. To build a long ladder, I'll need to purchase 15 to 20 years of expenses in long bonds. And speaking of the stock-like risk of long bonds, after losing 21% of its value in 2013, ZROZ gained 49% in 2014. PIMCO LTPZ, not limited to zero coupon bonds, fell 20% in 2013 and gained 20% in 2014, still a wild ride. Don't try to match long-duration liabilities with long-duration bond funds. Long bond funds have no place in a safe floor portfolio.

Many advisers make what I call the "mark-to-market" argument that funds and ladders are identical. This argument says that a ladder has the same volatility as the fund but that the ladder-holder simply ignores daily price volatility. That is correct, but the ladder offers the possibility of ignoring volatility by holding bonds to maturity while the fund does not, and if the investor is able to hold the bonds to maturity, that volatility is irrelevant.

While I am generally not swayed by this argument, its advocates do have a point. Even if I plan to hold all those bonds to maturity, there is a risk that I won't be able to, and this risk should be considered.

Retirees who are building a 4- or 5-year ladder to fund a gap or pay for college, for instance, are far less likely to be forced to sell bonds they intended to hold to maturity than are retirees who hold a 30-year ladder simply because there is less time for something to go wrong.

A retiree might be forced to sell bonds sooner than planned due to a financial crisis, such as a medical emergency, but there is also a significant risk that the bonds will be sold, not by the retiree, but by her estate or her heirs. I suppose this could be called "reverse longevity risk."

Longevity risk is the risk of outliving our savings. Buying a 30-year TIPS ladders and living 35 years would be an example of longevity risk. But, there is also a risk that a retiree might buy a 30-year ladder and live only 15 years. The bonds could then be sold at a loss by her estate if yields have risen, or by her young heirs who don't have a lot of need for a portfolio of long TIPS bonds at the age of 25.

Of course, should interest rates fall over time, the bonds might be sold before maturity at a profit, but I can live with that risk.

Put these three factors together and you see my concern: I buy a 30-year TIPS bond ladder today and lock in historically low interest rates. Rates rise for the next ten years, lowering the market value of my bonds, especially the long ones, and I die soon after that. The remaining bonds are inherited by my children, whose financial needs aren't well met by holding long TIPS bonds to maturity, so they sell them at a loss. Since the basis of these bonds is stepped up, they won't even get a tax break.

(I would suffer the same fate in this scenario if I funded those liabilities with a long TIPS bond fund instead of a ladder. So, this is also a concern with funding distant future liabilities with a bond fund.)

I have to weigh this risk of unplanned sales against the certainty, offered by a TIPS ladder held to maturity, of meeting future liabilities. Many factors would exacerbate or mitigate this risk. A married couple is much more likely to have at least one spouse who will survive long enough to use most of the ladder. The longer at least one spouse survives, the less likely the ladder will contain bonds with a large loss, because a bond's price will approach its face value as time passes.

Retirees with no bequest motive may care less about these risks than those who wish to leave an inheritance. (With no bequest motive, however, they might find a life annuity a better fit.) Retirees who have lots of other retirement income (over-savers) are less likely to need to sell bonds from their ladder in an emergency. Retirees who fund a lot of annual income with a ladder will have greater risk exposure than those that need only fund a small annual shortfall. The risk of needing to sell bonds before maturity varies significantly based on the household's individual situation.

Is there a way to fix this problem? Not a good one. We could use an annuity, but it will have even less liquidity than a ladder. A retiree who insists on following the daily market value of a TIPS fund should also want to follow the resale value of an annuity, and it will be even worse. The heirs of the TIPS bond ladder may see a loss, but the heirs of the retiree with an annuity will receive nothing at all.

Ultimately, I believe that Bernstein, Bodie, et al have it right. The safest way to provide certain future income is to purchase TIPS bonds and hold them to maturity. Yes, you may lock in low rates for a long time if you're unlucky, the strategy has opportunity cost and you might need to sell some bonds at a loss before they mature, but if your goal is strictly to provide income with certainty, this strategy is the best bet.

It is only when you add additional requirements, like a goal of maximizing yield or one of maximizing a bequest, or a concern about the market value of your assets should you have to sell them in a fire sale tomorrow, that the strategy shows some weaknesses. None of these are great objectives for a floor portfolio, by the way.

Still, on an individual basis, those additional requirements might be important to you and should be given consideration, especially for long ladders. They shouldn't be an issue for short ladders.

For most do-it-yourself retirees, I would summarize the last few posts on bond ladders and funds as follows. Retirees who aren't using bonds to match future liabilities will probably realize little advantage from buying a ladder instead of a fund. I believe TIPS ladders are the way to go for funding a few years, but using a short-duration fund for this purpose, instead, probably isn't a deal-breaker.

Long ladders and long bond funds are a different story. For the safest approach to providing certain income, ladders are the solution, especially if there is little risk that you will need to sell the bonds before maturity or if the residual value of the ladder isn't a concern. As I mentioned, long bond funds can be extremely volatile and do not belong in a safe floor portfolio.


P.S. Several readers asked after my recent discussion of bond funds versus ladders why I focussed on TIPS bond ladders. Wouldn't duration- and convexity-matching arguments also hold for funds and ladders of say, corporate bonds or munis?

They would, but Treasury bonds have no default risk so there is no need to diversify among issues. Other bonds, including corporates, do have default risk and we need to diversify among many issues for an acceptable level of safety. It would be difficult for most retirees to buy enough individual corporate bonds to adequately diversify, so mutual funds win over ladders in non-Treasury bond asset classes right off the bat based on their diversification advantage.


  1. Dick: You note that yields on TIPS are currently low. Isn't the associated interest rate risk somewhat mitigated by the "real" interest rate that TIPS provide? Historic real interest rates range up to about 3% (?), so the biggest hit you might take is an interest rate swing of about 2%, assuming the bonds are sold before maturity?

    1. I believe the long-term average real yield to maturity for TIPS bonds is just under 2%. If yields reverted to the mean from here, that would be about a 2% increase, but yields might well rise more than that.

      The "hit" that you will take if you sell the bond fund shares at the wrong time is a capital loss, which is approximated by the bond fund's average duration times the percent increase in rates.

      A short-duration fund might only have a loss of 5% if yields increase 2%. An intermediate TIPS fund could lose around 15% and a long zero fund like ZROZ could lose over 50% of its value.

      Thanks for writing!

    2. You can see the fund risk just by looking at a chart of TIP (click on "Maximum" to get the full pre-crash and crash performance).

      The transition from high inflation expectations to deflation from 2007 to 2008 meant that the fund plunged in value. 2013 saw another significant drop. It recovered over the next couple of years, but you don't want something with annual gyrations of 10% to 20% for your "safe" money to provide guaranteed income over the next couple of years.

      Because many stock market crashes coincide with significantly reduced inflation expectations, the TIPs funds will often be more correlated to equity funds than Treasury bond funds, so they make a good diversifier for bond funds, but not so good for equity funds.

  2. hi dirk, isn't a typical bond etf laddered theoratically. would they behave differently like bond funds in that those maturing bonds buys higher yielding bonds?

    1. A typical bond fund or ETF is not laddered, though there are a few, like PowerShares LDRI that are. Funds and ETFs are typically designed to maximize return, not to guarantee a certain amount of income on a certain date.

      Also, keep in mind that even a laddered ETF may not use the ladder you want. A ladder to fund a 5-year gap, for instance, needs to have a declining number of rungs while the laddered ETF will have a constant number of rungs, like 5, when the number you need might be 4 or 6 rungs declining to 1.

      No matter what kind of bond portfolio you have other than a fixed-length ladder of zero coupon bonds held to maturity, income from the coupon interest generated and the purchase of new bonds will be reinvested at a higher yield when rates increase. This higher yield will make up for the bond or bond fund's capital loss from the yield increase eventually if the bond or fund isn't sold. The approximate number of years required for a 1% change in yield to be recovered is the bond or fund's duration.

      It works like this. Say you purchase shares of LTPZ (described above) that has a duration of 20 at a cost of $10 per fund share when yields-to-maturity are about 0.5%. Yields increase 1% (to 1.5%). The fund's share price will immediately decline about 20% (its duration times the 1% increase) to $8.00 per share.

      Now, however, the fund is yielding 1.5% instead of 0.5%, so if yields remain unchanged from the new 1.5%, any interest generated or new bonds purchased will enjoy a higher yield. Eventually, this higher yield will make up for the $2 capital loss per share.

      This is all approximately correct, simplified for this explanation.

      Keep in mind that if we are selling from this fund each year instead of holding bonds to maturity, we add sequence of return risk. SOR risk is much higher for a long-duration bond fund than for a short-duration fund.

      Thanks for writing, Kyith.

    2. hi Drik, thanks for the explanation! i learn something new especially the theoratical correction for 1% of yield increase. i learn recently about sequence of return risk and u hit home the idea that this does not apply only to equity but also bonds. i guess my idea is bond funds distribute, but a major flaw in my thinking is that bond funds do not distribute, they reinvest and grow based on their own discretion.

      much of my experience with bond etf draws from my local singapore bond etf with a 5-6 year average duration. they do distribute but that is based on managers discretion, so i thought we could live off the distribution. i like the idea that capital is guaranteed based on the mandate, as i spoken to the sales director, and they stated they hold the 5-6 year bonds to maturityand during the last year sell it off (where value should approach par)

    3. That is not a typical fund. It is a "closed-end" fund. And, conceptually at least, it will be safer than a bond ETF that does not hold bonds to maturity. That, of course, is assuming the bonds are high quality to begin with. Otherwise, they could default and be worth nothing when they are redeemable.

  3. If I were to estimate my floor at $60K in today's dollars, 30 years from now, with Social Security covering an estimated $35K of that, would I just need to purchase a $25K 30-year TIP each year for the next 30 years? Or is it more complicated than that?

    1. Dave, it's a little more complicated than that. Instead of explaining in detail, I'll refer you to Wade Pfau's excellent and detailed blog on the subject, "How Do I Build a TIPS Bond Ladder for Retirement Income?"

      I'm a little confused by your question. Are you planning to retire 30 years from now, or retire now and fund retirement for 30 years?

  4. Helⅼo! Ꮤould yoᥙ mind if Ι share yⲟur blog ᴡith my facebook gгoup?

    Ƭhere's a lot of people that Ӏ thіnk would really appreciate yⲟur content.
    Please let me ҝnow. Thank ʏou