Is bond duration-matching uncontested for retirement?

[Poe22]

I've been wondering whether there's a consensus that the often discussed bond duration-matching is the optimal retirement strategy?

And is it suboptimal to implement it without TIPS, due to inflation risk?

[Dude2]

The idea in a nutshell is that when you match the duration of the fixed income instrument to the need for the money, then you have immunized yourself to interest rate risk. Essentially you will hold to maturity, and the bond will make good on the deal you had when you bought it. Extend the same logic to a fund. Whatever the economy did, interest rates did, etc. over the life of the bond is unimportant to you because the day you need the money will be the day you get exactly what you had coming.

It's a bit of an oversimplification that we can work things out so perfectly, but it helps to plan.

Now, fast forward to the part about the need for the money. What you really need is an equivalent purchasing power. Who cares that you will get $10,000 as agreed in 2035 when that $10,000 will not get you as much? TIPS will take care of that risk. Therefore, with TIPS as your bond, and laddering up the need, you can eliminate both interest rate risk and inflation risk.

Non-TIPS are higher risk, but those risks may not show up, providing higher reward. Take risk on the equity side. Bond side is for safety.

[NiceUnparticularMan]

Similar to the other poster, bonds can potentially serve two very different purposes.

In modern portfolio theory, the basic concept is you figure out which mix of risky assets has the best risk-adjusted expected returns, aka is most "efficient", and then you adjust your specific risk/return tradeoff with a varying proportion of a very low risk asset (sometimes called "risk free", but nothing is really entirely risk free).

In a standard lifecycle retirement model, investors are ultimately trying to provide a low-risk stream of real income from retirement until they pass away, but with some liquidity as well (in case they want to spend more for some reason at some point). But they also want to maximize the level of that income stream to the extent practical.

Investors in early accumulation start off with no very low risk assets, because their expected start for income is still far in the future and so they should focus on maximizing expected returns. But as income starts coming into the practical horizon, they gradually start accumulating very low risk assets to start satisfying the goal of providing for a low-risk stream of real income. Even at retirement, though, they have a long enough horizon that they will still have some significant percentage of retirement savings in the risky side of the portfolio.

Given this setup, the very low risk asset in question would be one which minimized the risk around that planned stream of real income with some liquidity. Ideally this would be something like an inflation-adjusted annuity that started at retirement but could also be partially liquidated on demand. No such thing exists today, but some combination of Social Security, possibly a pension, possibly the available annuities, and duration-matched IP bonds is going to get you as close as practical.

So yes, so far the very low risk part of a standard retirement savings and investment model is going to imply duration-matched IP bonds, and also possibly annuities, in addition to Social Security and whatever pension you might have.

However, there is still the other side of your portfolio, the risky side where your goal is to maximize risk-adjusted returns, aka efficiency. Diversified market portfolios of stocks are a very good start on an efficient portfolio, and in fact such a good start you could reasonably just end there. But, there are reasonable arguments for other things to add. Factor investing. Some REITS. Commodities futures have come and gone and come again. And so on.

And one other general class of thing you could consider is risky bonds. There is no consensus on which, if any, risky bonds you should consider on the risky side. But, LT government bonds have an argument. HIgh-quality bonds not from your currency zone have an argument. I've never seen a particularly good argument for corporate bonds, but some people believe that. And so on.

OK, so let's assume rather than your risky portfolio being 100% stocks, you think it should be something like 90% stocks, 10% some kind of risky bond. That could be reasonable, and deep in accumulation your portfolio could look like that. But as retirement approached, you would start phasing in your very low risk asset, which we determined should be duration-matched IP bonds (holding aside annuities). So maybe at retirement, you would be like 45% stocks, 5% risky bonds, 50% duration-matched IP bonds. Something like that.

OK, so in summary: I do think most models would agree the very low risk asset should be duration-matched IP bonds (plus maybe annuities and such). But there is no consensus AGAINST a small amount of risky bonds TOO. Those are just two very different purposes.

[petulant]

Yes, bond duration matching is contested. It is a powerful tool, but consider the following.

1. Duration matching a substantial part of the portfolio subjects that portfolio to volatility as interest rates move. Duration matching implies that the investor should be indifferent to bond volatility because they have appropriately developed the portfolio's duration. However, many investors do not feel this way because they also prefer to see stability in their portfolio; they are averse to paper losses. (In BH terms, the investor wants ballast.) Further, there may be uncertain, lumpy liquidity needs that cannot be perfectly timed on duration, meaning that the portfolio must also be evaluated for its short-term stability to meet surprise needs. Hence, both behaviorally and rationally, investors have reasons to temper long durations.

2. Stocks are very volatile and are meant to provide returns over a long period of time. These factors imply that stocks might have their own duration-like properties. Note that popular models for fundamental stock analysis include the discount cash flow model, or DCF model, in which all future returns from the stock are discounted to the present using a discount rate that effectively acts as a required or estimated rate of return. If interest rates across the economy fall very low, investors may start adjusting the discount rate used to estimate a stock's fundamental value, raising its value and hence its price. If rates rise, and if the discount rate then rises, then stocks may decline so that their expected return comes back into line with prevailing interest rates. A similar result obtains from CAPM if you allow that the relevant risk-free rate for some investors is driven by long-term rates rather than short-term rates.* If this is correct--that stocks have some kind of duration-like property--then whatever the investor's theoretical desire for duration exposure, that exposure could or should be reduced in recognition of the duration attributes in stocks. (This is not to say that stock volatility is only driven by duration properties. It is just saying that such properties might be one contributor to volatility.)

Putting those two points together, what could start as the perfect duration number may not be so perfect. For example, a 65-year-old retiree who prepares a bond ladder for a 30-year retirement with half their assets (the other half in stocks) might produce a portfolio with a duration like 10. However, by holding so much duration along with stocks, they expose themselves to volatility that they can't stomach. They also are concerned that if interest rates rise, their ability to withdraw from the portfolio to meet a spending shock (like a sudden new car, say) would be impacted (exacerbating sequence of returns risk). Hence, the investor might want to reduce the duration of their bonds. They could simply wrap them up in a bond fund that has a bit lower duration, like an intermediate Treasury bond fund. Or they could reduce the overall size of the bond ladder to a quarter of the portfolio, and they could hold the rest in a very-short-maturity bond fund, money market fund, or otherwise, reducing the average duration. Or they could buy an annuity with a quarter of the portfolio, which implicitly duration matches for them and then hides the volatility, while keeping the remaining quarter in (again) a short-maturity bond asset of some kind.

3. Legacy goals greatly complicate finding the theoretically perfect duration. For some investors (including many on BH), retirement assets are either greater than needed for retirement, or the total amount needed for retirement is uncertain such that it is likely, but not certain, some funds will be left over after the investor "graduates." Either way, the investor rationally maintains control and access over funds while alive in order to meet uncertain spending goals and then prepares a plan to leave the assets to beneficiaries and heirs, perhaps children, grandchildren, or charities. If duration matching is a species of liability matching, then the problem here is that the liabilities are contingent. At each point in time, the retiree walks down a foggy road with many forking paths, but they never know whether they are taking the fork of living longer with predictable spending needs or if they are taking the fork of graduating soon with unpredictable uses of their money having either shorter or longer duration. Hence, the investor is left doing their best, and analytical precision can be misleading.

* EDIT: Another mechanism for duration to impact stocks (rather than purely fundamental models) could come from short-term rates. Many investors participate in the stock market through leverage that ultimately relates to short-term rates, whether margin, futures, options, or otherwise. Long duration is tied to short short duration; in other words, borrowing at short-term rates effectively creates longer duration exposure, although with complications related to the yield curve. (Many of the threads on HFEA and otherwise develop this idea thoroughly, comparing leveraged Treasuries with shorter maturities to those with long maturities.) If the Federal Reserve raises short-term rates aggressively enough, the spread between margin rates and expected stock returns will decline, leading to leveraged investors selling their stocks, which creates a cycle where those short-term losses can pressure other leveraged investors to continue selling stocks. This cycle could be triggered even if few or none of the leveraged investors update robust fundamental models, since the interest rate increases can be tied to volatility, fears of recession, and so on; or because the investors care more about dividends covering interest costs than economic returns exceeding interest costs. The deleveraging process would continue until a new equilibrium where uneconomic leveraged investors have been flushed out and the expected return on stocks is commensurate with the interest rate environment. The resulting price behavior could be described as a duration-like property.

[dbr]

I think the above comments are appropriate. Duration matching is a theoretically well justified idea but is incomplete, complicated, and possibly undesirable in the context of actual portfolios for actual retirees. That does not mean one should not be aware of the mechanics that show how longer durations in bonds might not be appropriate and consider the implications when choosing bond investments.

Just my personal opinion and also how I choose to manage without attempting explicit liability matching.

[CloseEnough]

A potential complication to duration-matching in retirement, discussed somewhere at length in another thread, is that the "need for the money" from the fixed-income side of a portfolio may be difficult for a retiree to know with certainty, especially if they intend to draw down in a tax efficient manner that is correlated to their overall allocation.

[alex_686]

It is not.

The primary advantage is also its disadvantage. It attempts to eliminate all risk. Risk and return are linked, thus you are locking in low returns. When I was working with life insurance this was critical, but maybe not for the average retail investor.

[NiceUnparticularMan]

I agree with the above comments that uncertainty about the possibility you might want to spend sooner than planned implies you might want your fixed income to have a somewhat lower duration than if you could be certain about when you might spend, but conversely that a possible legacy motive could imply a longer duration (although giving while alive is nice). I similarly agree that stocks in effect act like very long bonds and therefore you again might want a somewhat lower duration in your fixed income to create a portfolio with the overall duration you are targeting--if the only goal of the portfolio is producing income while you live. If instead you are also planning to leave a legacy then maybe that is not a concern.

To me, though, all this is just sort of modifying the central concept of duration/liability matching given the complexities of a personal investor's possible liabilities (aka spending/giving).

Just one more complication--again bonds are not the only type of non-stock asset investors might have available, there is also Social Security, possibly pensions, annuities, unmarketed instruments like stable value funds and the TSP's G Fund, and so on. You can still do some sort of liability matching with all this stuff, but "duration" starts being a less and less useful way of thinking about liability matching the farther away you get from marketed bonds.

[martincmartin]

Bond duration matching always seemed like numerology to me.

If you back test with historical data, I believe bond duration matching doesn't do very well, intermediate term, or total bond market, is the "safest" option in terms of minimizing the chance of running out of money in retirement.

Why are you holding bonds?

Is it because when your investments go down you get worried and stressed out? Then use short term treasuries. This is Bill Bernstein's point.

Have you "won the game" and want a liability matching portfolio? Use a TIPS ladder. Just be aware of the problems with liability matching portfolios: you don't know your future liabilities, and you might outlive your ladder.

Do you want to minimize your chance of running out of money during retirement? Use total bond market in tax sheltered, something similar in taxable (plain treasuries, TIPS, and municipal is what I use).

[NiceUnparticularMan]

If you back test with historical data, I believe bond duration matching doesn't do very well, intermediate term, or total bond market, is the "safest" option in terms of minimizing the chance of running out of money in retirement.

Backtesting is not a reliable method generally, and moreover to my knowledge meaningful backtesting specifically with duration-matched TIPS is impossible because TIPS have not been around long enough.

Indeed, one specific scenario that has never been tested in the TIPS era is a prolonged inflationary stock crisis (the recent one is too short to count as prolonged). We know that such conditions were terrible for intermediate nominal bonds the last time we saw them, leading to the mid-late 60s being a terrible period for "safe withdrawal rates" for portfolios relying on such bonds. But since TIPS didn't exist back then, we can't do a "backtest" comparison.

[bobcat2]

Matching strategies whether with TIPS and Ibonds and/or with life annuities (real or graded) are for the amount of safe income you want in retirement beyond that supplied by Soc Sec and DB pension benefits. How much safe retirement income you want is up to you. All of these forms of income are hedging retirement income risk.

hedging - A method of transferring risk in which an action taken to reduce one's exposure to loss also causes one to give up possible gains.

BobK

[petulant]

When we say future liabilities aren't certain, it goes beyond household expenses and can apply to things like tax liabilities. Here's an example that some people might run into in the next few years, which could mess up a TIPS ladder. Imagine an investor used their entire traditional IRA at age 70 to build a TIPS ladder (leaving taxable accounts or Roth IRA for a stock allocation and any emergency fund). Let's say the investor was motivated to do so by historic real rates on TIPS, managing to get the entire ladder for a real interest rate of 1.5% and an overall duration of 10 or so. Within the last two years, the real interest rates were -0.5%. After three years, the investor must begin RMDs, and the duration has dropped a bit. However, due to a number of unexpected macroeconomic issues, real interest rates collapsed again to -.5%. This means that the market value of the TIPS ladder may increase by nearly 15%. (Consider if the duration had dropped to 7, then 7*2=14, which is only a rough estimate.) The investor should have been indifferent to the rate volatility because the need for real income is the same each year. However, the RMD calculation is now higher than expected. If real interest rates remain depressed at -.5% for several years, the investor will be required to take larger RMDs each year, paying more taxes than initially expected. The distributions may also cause higher taxation of social security benefits, further increasing the economic marginal tax rate on the RMDs. The investor will have to sell some of the TIPS ladder rungs inside the IRA to fulfill the RMDs. Further, even if assets can be shifted to rebuild the TIPS ladder in other accounts, there is a net loss in wealth from the additional taxation. The magnitude of this impact might be entirely manageable if the investor kept other assets, but if the vast majority of wealth was placed in the TIPS ladder, it might hurt. Other kinds of assets might have similar problems; I am not saying TIPS are bad. I am just saying that future liabilities are not perfectly predictable, and it can turn out that the perfect TIPS ladder wasn't as perfect as thought.

[Morik]

It is not.

The primary advantage is also its disadvantage. It attempts to eliminate all risk. Risk and return are linked, thus you are locking in low returns. When I was working with life insurance this was critical, but maybe not for the average retail investor.

I don't think this is correct. The risk that is eliminated by duration matching is not a compensated risk. As far as I know the risk materializes as greater dispersion of potential results (but with no increase in expected return).

E.g., you have liability in 20 years. You buy a 20 year bond--you know exactly how much money you will have in 20 years (in either nominal or inflation-adjusted terms, depending on the type of bond).
If you instead want to buy a 10 year bond and then buy another 10 year bond in 10 years, you have introduced uncertainty into how much money you will have in 20 years, but this uncertainty isn't compensated with a higher expected return AFAIK.

Similarly if you buy a 30 year bond and sell it in 20 years for your liability, again you have introduced uncertainty into how much money you'll have in 20 years, and I don't think that risk is compensated with higher expected return either.

[dcabler]

Matching strategies whether with TIPS and Ibonds and/or with life annuities (real or graded) are for the amount of safe income you want in retirement beyond that supplied by Soc Sec and DB pension benefits. How much safe retirement income you want is up to you. All of these forms of income are hedging retirement income risk.

hedging - A method of transferring risk in which an action taken to reduce one's exposure to loss also causes one to give up possible gains.

BobK

+100%
There is no requirement that it match your liabilities. Social security is also inflation adjusted, but those benefits know nothing about your liabilities. In other words, like so many other things, it depends on what you're comfortable with. I'm comfortable with X% of my cash flow having no interest rate risk and is inflation adjusted and I'm comfortable with Y% of my cash flow being generated by risky assets.

And I don't care about what happens to the value of my duration matched TIPS bonds either since what matters to me is the cash flow it generates, though that does require a mindset shift to start to ignore it. It's probably easier to ignore if you're holding actual bonds than a bond fund.

Cheers.

[alex_686]

It is not.

The primary advantage is also its disadvantage. It attempts to eliminate all risk. Risk and return are linked, thus you are locking in low returns. When I was working with life insurance this was critical, but maybe not for the average retail investor.

I don't think this is correct. The risk that is eliminated by duration matching is not a compensated risk. As far as I know the risk materializes as greater dispersion of potential results (but with no increase in expected return).

E.g., you have liability in 20 years. You buy a 20 year bond--you know exactly how much money you will have in 20 years (in either nominal or inflation-adjusted terms, depending on the type of bond).
If you instead want to buy a 10 year bond and then buy another 10 year bond in 10 years, you have introduced uncertainty into how much money you will have in 20 years, but this uncertainty isn't compensated with a higher expected return AFAIK.

Similarly if you buy a 30 year bond and sell it in 20 years for your liability, again you have introduced uncertainty into how much money you'll have in 20 years, and I don't think that risk is compensated with higher expected return either.

So only the first case is a true and correct example. At least in theory. Finding 10 year zero coupon Treasuries is not easy. 30 years are even harder.

So portfolio managers often have to bend to the realities of the world and use a combination of 10 year Treasuries and corporate bonds. There are actuary professionals who have thrown lots of math at the problem and have gad very good results. When was the last time you heard a life insurance company failing?

Critically you are using the wrong measuring stick, as my above bankruptcy comment alluded too. What is failure? Underperforming a different index with different risks then your goals? A wide dispersion of returns? No to both. It is about managing downside risk, or failure to meet the future cash flows.

[NiceUnparticularMan]

So portfolio managers often have to bend to the realities of the world and use a combination of 10 year Treasuries and corporate bonds. There are actuary professionals who have thrown lots of math at the problem and have gad very good results. When was the last time you heard a life insurance company failing?

When was the last time an insurance company offered an inflation-adjusted annuity?

Nominal annuities mean the risk of unexpected inflation remains on the annuitant, and so the insurance company is in no risk of failing by using nominal assets.

[secondopinion]

So portfolio managers often have to bend to the realities of the world and use a combination of 10 year Treasuries and corporate bonds. There are actuary professionals who have thrown lots of math at the problem and have gad very good results. When was the last time you heard a life insurance company failing?

When was the last time an insurance company offered an inflation-adjusted annuity?

Nominal annuities mean the risk of unexpected inflation remains on the annuitant, and so the insurance company is in no risk of failing by using nominal assets.

Correct. But the insurance company is not offering inflation-adjusted annuities, so what does it matter to them? Good point; objectives drive the portfolio design.

[toddthebod]

The biggest problem I have with the idea of duration-matching in retirement is that retirement is an uncertain length of time. Will you be retired for 5 years or 30 years?

Let's say you want to plan for 30 years, from age 65 to 95, and you set up a bond ladder, at what point do you start adding rungs for ages 96+? Or let's say you use a bond fund with the rule-of-thumb to match your duration to half your investment horizon, so 15 years. At age 70, are you still matching duration to age 80, or 85, or something in between? And again, when do you start planning for ages 96+?

[Kevin M]

<snip>
Finding 10 year zero coupon Treasuries is not easy. 30 years are even harder.

It's not at all hard to find them. The yields may not be attractive compared to coupon bonds, but they might be. Here are Treasury yields from yesterday:



The 10-year STRIPS yields are higher than the coupon bond yields. There is a large range of longer maturities where the STRIPS yield is higher than the coupon bond yield, but at the longest maturities, coupon bonds win.

For maturities from 10.1-year to 13.1-year there are STRIPS only--no coupon bonds.

The reason for the vertical red lines is that there is a stripped interest and stripped principal security for each maturity date, and the stripped interest security yields are higher.

Kevin

[alex_686]

So portfolio managers often have to bend to the realities of the world and use a combination of 10 year Treasuries and corporate bonds. There are actuary professionals who have thrown lots of math at the problem and have gad very good results. When was the last time you heard a life insurance company failing?

When was the last time an insurance company offered an inflation-adjusted annuity?

Nominal annuities mean the risk of unexpected inflation remains on the annuitant, and so the insurance company is in no risk of failing by using nominal assets.

Today? They are quite popular in the UK.

I am not sure if you are agreeing with me or not. My initial point was that duration-mating is not the gold standard. It is a standard and it does underpin a fair amount of risk methodology and thus portfolio construction theory. The reason why it is not the gold standard is that the base case is for worst case risk adverse portfolio.

Life insurance fits that. That nobody in the US wants a ultra-risk adverse portfolio does not undermine the theory or its uses.

[alex_686]

<snip>
Finding 10 year zero coupon Treasuries is not easy. 30 years are even harder.

It's not at all hard to find them. The yields may not be attractive compared to coupon bonds, but they might be. Here are Treasury yields from yesterday:

The 10-year STRIPS yields are higher than the coupon bond yields. There is a large range of longer maturities where the STRIPS yield is higher than the coupon bond yield, but at the longest maturities, coupon bonds win.

For maturities from 10.1-year to 13.1-year there are STRIPS only--no coupon bonds.

The reason for the vertical red lines is that there is a stripped interest and stripped principal security for each maturity date, and the stripped interest security yields are higher.

Kevin

I will modestly disagree with you on many points, but here are 2.

Coupon bonds have coupons thus they have a lower duration then STRIPS of the same maturity. If you were to decompose the cash flows of the coupon bonds you will get about the same yields as STRIPS, expect for the bid-ask spread, which is critical to my second point.

A STRIPS is actually a portfolio. A bank sets up a trust, dumps a bunch of Treasuries into that trust, divide up the cashflows, and charges a modest fee for that service. And for that you have to deal with a illiquid market. Why not as a insurance portfolio manager just buy the Treasuries and actively manage the risk of the coupon cash flows? You cut out the middleman, you get liquid treasuries, the risk is low, and the success rate is high.

Actually the standard is just to buy 10 year bonds. This is the most liquid tenor for long dated bonds and there is just not enough supply out there for the demand of the life insurance industry. I hear what you are saying but this is not how the world works in a practical sense.

[secondopinion]

However, there is still the other side of your portfolio, the risky side where your goal is to maximize risk-adjusted returns, aka efficiency. Diversified market portfolios of stocks are a very good start on an efficient portfolio, and in fact such a good start you could reasonably just end there. But, there are reasonable arguments for other things to add. Factor investing. Some REITS. Commodities futures have come and gone and come again. And so on.

And one other general class of thing you could consider is risky bonds. There is no consensus on which, if any, risky bonds you should consider on the risky side. But, LT government bonds have an argument. HIgh-quality bonds not from your currency zone have an argument. I've never seen a particularly good argument for corporate bonds, but some people believe that. And so on.

OK, so let's assume rather than your risky portfolio being 100% stocks, you think it should be something like 90% stocks, 10% some kind of risky bond. That could be reasonable, and deep in accumulation your portfolio could look like that. But as retirement approached, you would start phasing in your very low risk asset, which we determined should be duration-matched IP bonds (holding aside annuities). So maybe at retirement, you would be like 45% stocks, 5% risky bonds, 50% duration-matched IP bonds. Something like that.

OK, so in summary: I do think most models would agree the very low risk asset should be duration-matched IP bonds (plus maybe annuities and such). But there is no consensus AGAINST a small amount of risky bonds TOO. Those are just two very different purposes.

I think there is a lot of complexity to corporate bonds; from what I can analyze, one who tilt towards value will generally not benefit from corporate bonds since there is already some valuey nature to them: (https://www.portfoliovisualizer.com/opt ... ints=false).

On the flip side, increasing credit risk to speculative-grade does indeed make it behave a bit more growthy: (https://www.portfoliovisualizer.com/opt ... ints=false)

I cannot find a solid argument for or against them; corporate bonds put a percentage on what credit worthiness what do long-term. It is like fixing your possible return from stocks. Whether this is a good idea or not is debatable; I do not mind the risky nature here, but I am guilty of being a being of a market timer with them as well (more of a value timer in the sense I am more concerned with the realized long-term gain than the direction of the present market).

[NiceUnparticularMan]

That nobody in the US wants a ultra-risk adverse portfolio does not undermine the theory or its uses.

The last time one of these conversations (that I participated in at least) looked at the history of CPI-adjusted annuities in the US, my assumption going in was they disappeared because they were more transparent about their low real rates of return than nominal annuities. But it turns out it looked like CPI-adjusted annuities had a much higher "premium" charge than TIPS versus nominal Treasuries (where the premium charge is so low as to be negligible).

So I am now not so sure it was that people stopped wanting CPI-adjusted annuities per se, as that the normal investment practices of insurance companies could not offer them competitively.

Of course this isn't really about duration matching. It is about the real/nominal part of liability matching. But in terms of timing, insurance companies are setting the terms of their annuities and pooling them in ways that make their schedule of nominal liabilities far, far more certain than an individual's preferred retirement spending schedule will end up being.

[NiceUnparticularMan]

I cannot find a solid argument for or against them

I was more referring to the lack of a solid argument for them, and generally I am fine on the risky side with only including something besides stocks if there is a sufficiently strong positive case.

But that said, most of the efficient portfolio models I have seen have tended to end up with very little if anything in corporate bonds in your own currency as part of the efficient risky portfolio. Basically, they typically end up dominated by some combination of stocks, maybe longer-term government bonds in your currency, and/or maybe bonds not in your own currency.

That said, these sorts of models tend to use just some boring index of investment grade corporates. So they can't rule out the idea there may be some fancier way of using some corporate bonds in your own currency to improve portfolio efficiency. I just have zero interest in trying to figure such a thing out.

[secondopinion]

I cannot find a solid argument for or against them

I was more referring to the lack of a solid argument for them, and generally I am fine on the risky side with only including something besides stocks if there is a sufficiently strong positive case.

But that said, most of the efficient portfolio models I have seen have tended to end up with very little if anything in corporate bonds in your own currency as part of the efficient risky portfolio. Basically, they typically end up dominated by some combination of stocks, maybe longer-term government bonds in your currency, and/or maybe bonds not in your own currency.

That said, these sorts of models tend to use just some boring index of investment grade corporates. So they can't rule out the idea there may be some fancier way of using some corporate bonds in your own currency to improve portfolio efficiency. I just have zero interest in trying to figure such a thing out.

Efficient in what way? I am not sold on the Sharpe Ratio because it is a short-term metric and ignores skew. Is there an analysis that considers longer-term metrics?

[petulant]

The tricky thing about "efficient portfolios" is that most of the theories regarding them involve assumptions like homogenous investors with no market restrictions or tax differences. If you run an efficient frontier analysis on PortfolioVisualizer with no allowed stock allocation but only various fixed income assets as options, then corporate bonds will feature in pretty much every efficient portfolio, sometimes as the vast majority of the portfolio. Since banks and insurance companies can hold long-term bonds under historical cost accounting and "hide" the volatility, but have much more stringent restrictions on equity holdings, then they are natural buyers for corporate bonds in ways that individual investors are not. It's entirely possible and fair that things that work for other types of investors don't make sense for relatively small retiree investors.

[hudson]

I've been wondering whether there's a consensus that the often discussed bond duration-matching is the optimal retirement strategy?

And is it suboptimal to implement it without TIPS, due to inflation risk?

I don't know if it's optimal for everyone, but it fits my situation.
Why? Inflation is a real threat. I could rollover short treasuries for the duration or duration match 70% TIPS and 30% nominals. I'm all fixed so stocks can't save me.

Suboptimal to implement without TIPS?
For me, yes.
One could roll short term nominal treasuries for the long term. That might win, but I can't warm up to that.
Duration matched TIPS are guaranteed to do the job.

Before I read about duration matching, I was long bond shy; no more.

[alex_686]

That nobody in the US wants a ultra-risk adverse portfolio does not undermine the theory or its uses.

The last time one of these conversations (that I participated in at least) looked at the history of CPI-adjusted annuities in the US, my assumption going in was they disappeared because they were more transparent about their low real rates of return than nominal annuities. But it turns out it looked like CPI-adjusted annuities had a much higher "premium" charge than TIPS versus nominal Treasuries (where the premium charge is so low as to be negligible).

So I am now not so sure it was that people stopped wanting CPI-adjusted annuities per se, as that the normal investment practices of insurance companies could not offer them competitively.

So here is what killed the market.

Half of it was the bond market. Corporates stopped offering long dated floating rate bonds which do a decent job of inflation protection. Insurance companies could not transition to TIPS because the Treasury was only offering a very limited supply. There were real supply issues back 20 years ago. Even 10. There were regulatory reasons for this. Pension funds which offered CPI adjustments got a actuarial bonus if they held TIPS. So they just hovered up the whole market.

Half of it was about the consumers. Inflation had fallen to a low and steady rate. Consumers kind of forgot about it and were not willing to pay for that type of protection. We will see if this spike in inflation changes things.

Once again I will contrast this with the UK system. They issue sufficient inflation linked gilts and stimulate demand with their retirement regulations. There it works just fine.

[petulant]

That nobody in the US wants a ultra-risk adverse portfolio does not undermine the theory or its uses.

The last time one of these conversations (that I participated in at least) looked at the history of CPI-adjusted annuities in the US, my assumption going in was they disappeared because they were more transparent about their low real rates of return than nominal annuities. But it turns out it looked like CPI-adjusted annuities had a much higher "premium" charge than TIPS versus nominal Treasuries (where the premium charge is so low as to be negligible).

So I am now not so sure it was that people stopped wanting CPI-adjusted annuities per se, as that the normal investment practices of insurance companies could not offer them competitively.

So here is what killed the market.

Half of it was the bond market. Corporates stopped offering long dated floating rate bonds which do a decent job of inflation protection. Insurance companies could not transition to TIPS because the Treasury was only offering a very limited supply. There were real supply issues back 20 years ago. Even 10. There were regulatory reasons for this. Pension funds which offered CPI adjustments got a actuarial bonus if they held TIPS. So they just hovered up the whole market.

Half of it was about the consumers. Inflation had fallen to a low and steady rate. Consumers kind of forgot about it and were not willing to pay for that type of protection. We will see if this spike in inflation changes things.

Once again I will contrast this with the UK system. They issue sufficient inflation linked gilts and stimulate demand with their retirement regulations. There it works just fine.

Somebody, I don't recall who, regularly claims that insurance companies could have just used CPI derivatives. They claimed that there is a large, low-friction, robust market for these derivatives. So, they blame the whole problem on consumer demand. What is your take on that? I was skeptical that a derivatives market would actually work since the counterparties are going to hedge somewhere, meaning a large enough derivatives market would still require the TIPS or similar asset supply. I also can't find anything like CPI futures on the cmegroup.com website, which makes me wonder how big the market would be.

[alex_686]

Somebody, I don't recall who, regularly claims that insurance companies could have just used CPI derivatives. They claimed that there is a large, low-friction, robust market for these derivatives. So, they blame the whole problem on consumer demand. What is your take on that? I was skeptical that a derivatives market would actually work since the counterparties are going to hedge somewhere, meaning a large enough derivatives market would still require the TIPS or similar asset supply. I also can't find anything like CPI futures on the cmegroup.com website, which makes me wonder how big the market would be.

Not futures, but I can confirm that there is a large robust CPI swap market. Swaps are just a collection of forwards, and futures are just a special type of forward. I suspect that opening positions for these swaps start at 5m. As such they are limited to institutional players. Large enough that the contracts are standardized so they can be centrally cleared (CCP) through the London Clearing House (IIRC). For those who work with such things that is kind of the gold standard. Many swaps don't have the necessary volume or notional balance to warrant a CCP. I have never worked directly with these swaps.

Note, by the time standardized inflation swaps came about the Treasury was issuing many more TIPS. IIRC the expected swap CPI inflation rate is lower than the TIPS spread.

But just because there is a large market doesn't mean it is large enough. Insurance companies would be a natural party to one side of the swap, but I don't know who would be a logical counterparty to the other. Then again, I have not thought about it that much.

You could go to the Dep. of Ag and figure out volumes and outstanding notional. It gets reported on.

[NiceUnparticularMan]

I cannot find a solid argument for or against them

I was more referring to the lack of a solid argument for them, and generally I am fine on the risky side with only including something besides stocks if there is a sufficiently strong positive case.

But that said, most of the efficient portfolio models I have seen have tended to end up with very little if anything in corporate bonds in your own currency as part of the efficient risky portfolio. Basically, they typically end up dominated by some combination of stocks, maybe longer-term government bonds in your currency, and/or maybe bonds not in your own currency.

That said, these sorts of models tend to use just some boring index of investment grade corporates. So they can't rule out the idea there may be some fancier way of using some corporate bonds in your own currency to improve portfolio efficiency. I just have zero interest in trying to figure such a thing out.

Efficient in what way? I am not sold on the Sharpe Ratio because it is a short-term metric and ignores skew. Is there an analysis that considers longer-term metrics?

So this is a paper I like which does various robustness tests in addition to a normal mean-variance evaluation:

https://www.econstor.eu/bitstream/10419 ... 710937.pdf

It is actually mostly about stable value funds, but they look at a lot of portfolios assuming no stable value fund is available, and generally corporate bonds end up dominated anyway. There is an exception--in the 2002 to 2010 period, corporate bonds had a much larger role. But not outside that period, and they explain that at least in part by suggesting there was a unique departure between government and corporate bonds in that period.

Anyway, again I don't want to say any of this is remotely conclusive. It is more just suggestive that the case for not including corporate bonds--at least in this way--is mildly stronger than the case for including them.

[NiceUnparticularMan]

The tricky thing about "efficient portfolios" is that most of the theories regarding them involve assumptions like homogenous investors with no market restrictions or tax differences. If you run an efficient frontier analysis on PortfolioVisualizer with no allowed stock allocation but only various fixed income assets as options, then corporate bonds will feature in pretty much every efficient portfolio, sometimes as the vast majority of the portfolio. Since banks and insurance companies can hold long-term bonds under historical cost accounting and "hide" the volatility, but have much more stringent restrictions on equity holdings, then they are natural buyers for corporate bonds in ways that individual investors are not. It's entirely possible and fair that things that work for other types of investors don't make sense for relatively small retiree investors.

Yeah, there is obviously a large market for corporates, but it is worth remembering many of those entities buying corporates will be government, institutions, and other corporations. They may well be matching liabilities or satisfying other investment goals in ways that make sense for them, but are inapplicable to me.

Personally, I feel similarly about many derivatives, which also have a large market. There are all sorts of derivatives that make perfect sense for specific sorts of entities, but I have zero interest in investing in most derivatives.

[Kevin M]

<snip>
Finding 10 year zero coupon Treasuries is not easy. 30 years are even harder.

It's not at all hard to find them. The yields may not be attractive compared to coupon bonds, but they might be. Here are Treasury yields from yesterday:

The 10-year STRIPS yields are higher than the coupon bond yields. There is a large range of longer maturities where the STRIPS yield is higher than the coupon bond yield, but at the longest maturities, coupon bonds win.

For maturities from 10.1-year to 13.1-year there are STRIPS only--no coupon bonds.

The reason for the vertical red lines is that there is a stripped interest and stripped principal security for each maturity date, and the stripped interest security yields are higher.

Kevin

I will modestly disagree with you on many points, but here are 2.

Coupon bonds have coupons thus they have a lower duration then STRIPS of the same maturity. If you were to decompose the cash flows of the coupon bonds you will get about the same yields as STRIPS, expect for the bid-ask spread, which is critical to my second point.

A STRIPS is actually a portfolio. A bank sets up a trust, dumps a bunch of Treasuries into that trust, divide up the cashflows, and charges a modest fee for that service. And for that you have to deal with a illiquid market. Why not as a insurance portfolio manager just buy the Treasuries and actively manage the risk of the coupon cash flows? You cut out the middleman, you get liquid treasuries, the risk is low, and the success rate is high.

Actually the standard is just to buy 10 year bonds. This is the most liquid tenor for long dated bonds and there is just not enough supply out there for the demand of the life insurance industry. I hear what you are saying but this is not how the world works in a practical sense.

You did not disagree with a single point I made! I simply pointed out that your statement about the difficulty of "finding" zero-coupon bonds of specific maturities was incorrect; STRIPS are zero-coupon bonds.

I did not say there weren't advantages of using coupon bonds instead--I explicitly acknowledged that there might be (unacceptable yields). Of course there are many other reasons to prefer one over the other, but I didn't say there weren't.

I have read the textbook entries on immunization strategies, including discussions of issues with zeros that lead bond portfolio managers to include coupon bonds. That's actually a big part of the motivation of duration matching rather than cash flow matching. Cash flow matching is easier with zeros, but due to issues with liquidity or whatever, a portfolio manager might prefer coupon bonds, and thus might prefer a duration-matching strategy.

[NiceUnparticularMan]

Once again I will contrast this with the UK system. They issue sufficient inflation linked gilts and stimulate demand with their retirement regulations. There it works just fine.

And I suspect we are not contradicting each other here, but my takeaway from this is the fact that US insurance companies are not offering CPI-adjusted annuities is not so much that those are inherently undesirable, as that unlike UK insurance companies they lack the necessary supply of CPI-adjusted underlying assets, and trying to simulate CPI-adjustments with the available nominal assets is too inefficient to create a competitive product.

But a personal USD investor can freely invest in TIPS up to any reasonable scale, so doesn't have the same problem matching real liabilities.

[petulant]

Somebody, I don't recall who, regularly claims that insurance companies could have just used CPI derivatives. They claimed that there is a large, low-friction, robust market for these derivatives. So, they blame the whole problem on consumer demand. What is your take on that? I was skeptical that a derivatives market would actually work since the counterparties are going to hedge somewhere, meaning a large enough derivatives market would still require the TIPS or similar asset supply. I also can't find anything like CPI futures on the cmegroup.com website, which makes me wonder how big the market would be.

Not futures, but I can confirm that there is a large robust CPI swap market. Swaps are just a collection of forwards, and futures are just a special type of forward. I suspect that opening positions for these swaps start at 5m. As such they are limited to institutional players. Large enough that the contracts are standardized so they can be centrally cleared (CCP) through the London Clearing House (IIRC). For those who work with such things that is kind of the gold standard. Many swaps don't have the necessary volume or notional balance to warrant a CCP. I have never worked directly with these swaps.

Note, by the time standardized inflation swaps came about the Treasury was issuing many more TIPS. IIRC the expected swap CPI inflation rate is lower than the TIPS spread.

But just because there is a large market doesn't mean it is large enough. Insurance companies would be a natural party to one side of the swap, but I don't know who would be a logical counterparty to the other. Then again, I have not thought about it that much.

You could go to the Dep. of Ag and figure out volumes and outstanding notional. It gets reported on.

Thanks. That's informative. Your comment on counterparty ties to what I was getting at--with no natural counterparty, the other side is going to have to hedge the inflation exposure, which would require TIPS or another asset bundle.

I didn't find anything on the Ag website but I found the following report from CFTC. It doesn't seem to break out inflation swaps directly; they are included in "Other." I see there are only $10 trillion in notional USD swaps for "Other." If inflation swaps are only $1-2 trillion of that, I would think one insurance company would be able to sustain a CPI annuity, but maybe not the whole industry. I imagine another problem would be getting the tenor right. Insurance companies would probably want 30-year tenors, which might be a fraction of the market.

https://www.cftc.gov/MarketReports/Swap ... ssExp.html

[alex_686]

Thanks. That's informative. Your comment on counterparty ties to what I was getting at--with no natural counterparty, the other side is going to have to hedge the inflation exposure, which would require TIPS or another asset bundle.

I didn't find anything on the Ag website but I found the following report from CFTC. It doesn't seem to break out inflation swaps directly; they are included in "Other." I see there are only $10 trillion in notional USD swaps for "Other." If inflation swaps are only $1-2 trillion of that, I would think one insurance company would be able to sustain a CPI annuity, but maybe not the whole industry. I imagine another problem would be getting the tenor right. Insurance companies would probably want 30-year tenors, which might be a fraction of the market.

https://www.cftc.gov/MarketReports/Swap ... ssExp.html

I stand corrected. The CTFC has been spun off from the Dep of Ag. Still is overseen by the congressional sub-committee on agriculture. So that fight is still not done.

And here is something random that I found on the web saying there is only 16.4b outstanding in Oct 2022.
https://www.risk.net/investing/derivati ... standstill

And it looks like the swap market is much more robust for UK and EU inflation swaps. It is large enough to run a few inflation hedged corporate bond ETFs.

[Poe22]

I am grateful and admittedly overwhelmed by all the valuable responses! I believe I've gained a much better understanding of the pros and cons of duration matching, thanks to you all!

Frankly, I'm pretty new to investing and do not have any finance or even "numbers" background, so I guess I only got half of your points, but I hope the most important ones.

At age 40 with only 25x annual expenses and planning for a 50 year retirement, I guess I'll need some income from the bond-side too, so that would make duration matching suboptimal for me I guess? Too risk-averse and therefore too little returns for a prolonged retirement?

[alex_686]

I am grateful and admittedly overwhelmed by all the valuable responses! I believe I've gained a much better understanding of the pros and cons of duration matching, thanks to you all!

Frankly, I'm pretty new to investing and do not have any finance or even "numbers" background, so I guess I only got half of your points, but I hope the most important ones.

At age 40 with only 25x annual expenses and planning for a 50 year retirement, I guess I'll need some income from the bond-side too, so that would make duration matching suboptimal for me I guess? Too risk-averse and therefore too little returns for a prolonged retirement?

I would suggest that you abandoned this approach.

It is a workable strategy. You have psudo-liabilities on one side (required retirement spending) and assets on the other. You match these together and immunize your risk.

But you are not going to do that. You are going to have one generalized pot of funds to cover retirement. In one sense this is good news. It allows a higher level of integration between time periods, which allows for a more efficient portfolio. i.e., higher returns for lower risk. The better news is that lots of math can help you solve this problem. Or maybe this is just good news for those of us who like numbers. It is a more complex and abstract approach.

Next, how risk adverse are you? 25x is a bit light for 50 years. I mean, there is a good 50% chance you will make it, and about 1/3 you will run out after 35 years or so. (ballpark numbers going off of gut instinct)

[Poe22]

Next, how risk adverse are you? 25x is a bit light for 50 years. I mean, there is a good 50% chance you will make it, and about 1/3 you will run out after 35 years or so. (ballpark numbers going off of gut instinct)

Not risk adverse at all, I sleep tight even in bear markets

So you'd suggest a classic 60/40 (non duration-matched) would be more promising for an early retirement with modest net worth?

[BrooklynInvest]

I will readily admit to not understanding the rationale for duration matching at all - for individuals. Would welcome someone correcting my thinking!

I get it for an institution that may have a fixed payment at a certain future point and wants to squeeze every drop of yield out of their receivables before that date. But for an individual with an unknown "expiration date", I'm matching the duration of my bond portfolio to my life expectancy to eliminate rate risk as I get closer to death. That right?

Using me as an example - If I'm 55 my life expectancy is, say 85 so my duration should be a whopping 30 years? Versus the Agg's 6-7 years? So I massively extend my duration/volatility now and then each few years shorten it based on my reduced life expectancy? I'm eventually 65 and my life expectancy is say 82 so I reduce my duration from 30 to a still lengthy (and therefore volatile) 17?

And so on. At 80 say my life expectancy is 85, now I'm at 5. I'm 100, let's say my life expectancy/duration is 2.

To implement I would bail on total bond and extend my duration/volatility out to 30 today, and reduce it to zero over the length of my retirement. So I take on a huge amount of rate risk today (the yield on a 5 year bond is similar to that of a 30 but the vol increases significantly at the far end of the curve) and gradually eliminating all yield above that of cash.

I like yield. Hell, if I live a very long time I might really need yield.

Also, what do I do with my stocks? Surely by far the largest source of portfolio volatility is my stocks, which I've still got even though I've eliminated rate risk in my bonds? Or do I eliminate equity risk as well going from 60-40-0 to 0-0-100 over time?

[alex_686]

I will readily admit to not understanding the rationale for duration matching at all - for individuals. Would welcome someone correcting my thinking!

I get it for an institution that may have a fixed payment at a certain future point and wants to squeeze every drop of yield out of their receivables before that date. But for an individual with an unknown "expiration date", I'm matching the duration of my bond portfolio to my life expectancy to eliminate rate risk as I get closer to death. That right?

Using me as an example - If I'm 55 my life expectancy is, say 85 so my duration should be a whopping 30 years? Versus the Agg's 6-7 years? So I massively extend my duration/volatility now and then each few years shorten it based on my reduced life expectancy? I'm eventually 65 and my life expectancy is say 82 so I reduce my duration from 30 to a still lengthy (and therefore volatile) 17?

And so on. At 80 say my life expectancy is 85, now I'm at 5. I'm 100, let's say my life expectancy/duration is 2.

To implement I would bail on total bond and extend my duration/volatility out to 30 today, and reduce it to zero over the length of my retirement. So I take on a huge amount of rate risk today (the yield on a 5 year bond is similar to that of a 30 but the vol increases significantly at the far end of the curve) and gradually eliminating all yield above that of cash.

I like yield. Hell, if I live a very long time I might really need yield.

Also, what do I do with my stocks? Surely by far the largest source of portfolio volatility is my stocks, which I've still got even though I've eliminated rate risk in my bonds? Or do I eliminate equity risk as well going from 60-40-0 to 0-0-100 over time?

How much actuarial math do you want to throw at the problem? Let say you figure you have a 50% of being alive in 30 years. Figure out how much cashflow you need for that year, multiply it by 50%, discount by 30 years. Bingo!

On that, there is no place of equities - that is just not in the framework.

I am a modest protonate of Constant Proportion Portfolio Investment (CPPI). You can use this framework and duration matching to figure out what a minimum sized portfolio is to meet your required goals. If you are above you can keep assets in higher risk and higher return assets and/or spend on desired or stretch goals. As your portfolio falls to that CPPI you shift assets towards bonds and/or reduce spending to required levels.

[Horton]

Duration matching is a sensible way to fund expenses in retirement.

For 50 years though duration matching won’t work - there’s no guarantees when your planning a “retirement” that long.

[Peter G]

Using me as an example - If I'm 55 my life expectancy is, say 85 so my duration should be a whopping 30 years? Versus the Agg's 6-7 years? So I massively extend my duration/volatility now and then each few years shorten it based on my reduced life expectancy? I'm eventually 65 and my life expectancy is say 82 so I reduce my duration from 30 to a still lengthy (and therefore volatile) 17?

Don’t think so. This was covered recently. Your first duration is not 30 years but ~30/2. When you get to 65 years old, it’s dropped to 17/2.
viewtopic.php?p=7058849#p7058849

[WoodSpinner]

I will readily admit to not understanding the rationale for duration matching at all - for individuals. Would welcome someone correcting my thinking!

I get it for an institution that may have a fixed payment at a certain future point and wants to squeeze every drop of yield out of their receivables before that date. But for an individual with an unknown "expiration date", I'm matching the duration of my bond portfolio to my life expectancy to eliminate rate risk as I get closer to death. That right?

Using me as an example - If I'm 55 my life expectancy is, say 85 so my duration should be a whopping 30 years? Versus the Agg's 6-7 years? So I massively extend my duration/volatility now and then each few years shorten it based on my reduced life expectancy? I'm eventually 65 and my life expectancy is say 82 so I reduce my duration from 30 to a still lengthy (and therefore volatile) 17?

And so on. At 80 say my life expectancy is 85, now I'm at 5. I'm 100, let's say my life expectancy/duration is 2.

To implement I would bail on total bond and extend my duration/volatility out to 30 today, and reduce it to zero over the length of my retirement. So I take on a huge amount of rate risk today (the yield on a 5 year bond is similar to that of a 30 but the vol increases significantly at the far end of the curve) and gradually eliminating all yield above that of cash.

I like yield. Hell, if I live a very long time I might really need yield.

Also, what do I do with my stocks? Surely by far the largest source of portfolio volatility is my stocks, which I've still got even though I've eliminated rate risk in my bonds? Or do I eliminate equity risk as well going from 60-40-0 to 0-0-100 over time?

I will use myself as an example …

I need significant Cashflow for the next 6 years to meet my Retirement Goals. Once I start SS my Cashflow needs become minimal ( under 2%).

So duration matching is a very useful approach for positioning my assets to meet my Cashflow needs. Plus it’s not the entire portfolio so there are assets remaining for other Retirement or Estate Planning Goals.

You don’t have to duration match for your expected lifespan — I focus on a rolling 10 year period….

WoodSpinner

[Dude2]

I will readily admit to not understanding the rationale for duration matching at all - for individuals. Would welcome someone correcting my thinking!

I get it for an institution that may have a fixed payment at a certain future point and wants to squeeze every drop of yield out of their receivables before that date. But for an individual with an unknown "expiration date", I'm matching the duration of my bond portfolio to my life expectancy to eliminate rate risk as I get closer to death. That right?

Using me as an example - If I'm 55 my life expectancy is, say 85 so my duration should be a whopping 30 years? Versus the Agg's 6-7 years? So I massively extend my duration/volatility now and then each few years shorten it based on my reduced life expectancy? I'm eventually 65 and my life expectancy is say 82 so I reduce my duration from 30 to a still lengthy (and therefore volatile) 17?

And so on. At 80 say my life expectancy is 85, now I'm at 5. I'm 100, let's say my life expectancy/duration is 2.

To implement I would bail on total bond and extend my duration/volatility out to 30 today, and reduce it to zero over the length of my retirement. So I take on a huge amount of rate risk today (the yield on a 5 year bond is similar to that of a 30 but the vol increases significantly at the far end of the curve) and gradually eliminating all yield above that of cash.

I like yield. Hell, if I live a very long time I might really need yield.

Also, what do I do with my stocks? Surely by far the largest source of portfolio volatility is my stocks, which I've still got even though I've eliminated rate risk in my bonds? Or do I eliminate equity risk as well going from 60-40-0 to 0-0-100 over time?

These questions are fundamental, and people should have a plan of some sort. Remember what problem you are trying to solve (income in retirement). Sounds like you are taking an approach of a basket of risk assets, and then perhaps after accumulating a large pot you will draw down at some safe withdrawal rate, hoping the pot is large enough to support that. Sound right? Theoretically if the safe withdrawal rate is less than the dividend rate of stocks, your money could last forever. Then in this model, I don't see why people wouldn't just plan to be 100% stocks period.

On another hand, you have people that wish to invest in bonds, and they wish to do so smartly. Perhaps these people are worried about the volatility of stocks and there being no guarantees. Also the size of the pot fluctuates, so how do you know exactly when your pot is large enough that you can pull the plug and retire?

The bond investor surely wants to insulate themselves from interest rate risk, and "matching duration to need" sure seems like a valid approach. You aren't mentioning that there is overall portfolio duration, but you also can have various maturity buckets. I can still keep an overall long portfolio duration and yet have bonds maturing annually. Consider the poor man's liability matching portfolio where I buy $10k in IBonds every year. After 30 years, they will start to mature. If I don't need the money I can buy another 10K IBond with it. Eventually, when I retire, I end up with a 30 year "pension" that pays me $10k real each year added to my social security. Extend this simple example to construction of a bond ladder of any type that will provide you income annually.

I agree that it is difficult to buy long bonds and feel comfortable with it. (The fear being having to deviate from the plan at the wrong time, i.e. cash out when long bonds have dropped greatly due to interest rate movements). The yield curve slope should be a factor in going long. The presumption is that it is "worth it" to go out in duration, but I say this depends on slope. TIPS are the only kind I do long, but we've seen times where you can't buy them at a positive real rate, making it expensive to pay for safety. I think intermediate term is something we can practically live with, but you also have some assets in shorter duration funds. You can move a year's expenses from intermediate to shorter annually, from shorter to MM annually. The immunity from interest rate risk is just as mitigated, I think, and you are pulling your income as required. Does that make sense at all? I think Vineviz should point out the folly of my simplicity at this point. Bonds are just math, and it isn't ideal or rational to neglect benefits of long bonds and take on reinvestment risk when need for money is far out. Shorter duration bonds are just as affected by everything as long bonds, etc. I agree. Sometimes things are behavioral.

What you may be getting at is how to reconcile the classic 60/40 portfolio to the two extremes of "pure stock" vs "liability matching income stream". It may be that the 60/40 camp still wants to consider their portfolio a large pot to draw from at some safe withdrawal rate, so they still need their pot sufficiently large. They just have a higher confidence in the value and the stability versus with pure stock. They will ratchet up their bond percentage closer to retirement (for increased stability), but never going outside the 75/25 to 25/75 range. They may also keep a longer overall duration but with a couple of buckets of maturity and moving money around annually so that they are insulated from interest rate risk as well. Maybe that addresses the question of what you do with bonds in terms of duration matching inside your 60/40 (or whatever percentage) portfolio. Wouldn't you still want to match duration to need?

[NiceUnparticularMan]

To implement I would bail on total bond and extend my duration/volatility out to 30 today, and reduce it to zero over the length of my retirement. So I take on a huge amount of rate risk today (the yield on a 5 year bond is similar to that of a 30 but the vol increases significantly at the far end of the curve) and gradually eliminating all yield above that of cash.

I like yield. Hell, if I live a very long time I might really need yield.

OK, so a few necessary modifications.

First, as another poster pointed out, you are really looking at a ladder of IP bonds over your expected lifespan, not one IP bond that matures at your expected death. And that would have a duration around half of your remaining lifespan.

Second, you do in fact have to have some plan for longevity risk, meaning living longer than expected. That is why ideally you would be looking not at IP bonds, but inflation adjusted annuities. Fortunately, US investors typically have one of those in the form of Social Security. Unfortunately, these days they can't buy any more.

OK, so what you really have to try to do is synthesize an inflation adjusted annuity that would dovetail with Social Security. And long story short, that can roughly be done with a combination of TIPS, stocks, and possibly some of the available nominal annuities. And liability-matching your TIPS will likely make that a closer match for an actual inflation adjusted annuity.

That's going to be about as robust an approach for turning $N into a real income stream of $X annually over your remaining lifespan as you can get. If you look at what, say, DFA does with its Target funds, that is basically what they are doing (minus the annuities, which they could not offer in a Target fund). And you can try something different to get a bigger $X for the same $N, but likely what you would try will increase your risk of failure in some notable way.

Now, if you want to also invest for other purposes, you don't have to put all of your investments into this approach. If you have enough extra to invest and your are flexible about exactly what is used for what purpose, you could do something entirely different.

But if you are interested in maximizing the "safe" $X you can get for $N, that will be a good way to do it.

[BrooklynInvest]

I will readily admit to not understanding the rationale for duration matching at all - for individuals. Would welcome someone correcting my thinking!

I get it for an institution that may have a fixed payment at a certain future point and wants to squeeze every drop of yield out of their receivables before that date. But for an individual with an unknown "expiration date", I'm matching the duration of my bond portfolio to my life expectancy to eliminate rate risk as I get closer to death. That right?

Using me as an example - If I'm 55 my life expectancy is, say 85 so my duration should be a whopping 30 years? Versus the Agg's 6-7 years? So I massively extend my duration/volatility now and then each few years shorten it based on my reduced life expectancy? I'm eventually 65 and my life expectancy is say 82 so I reduce my duration from 30 to a still lengthy (and therefore volatile) 17?

And so on. At 80 say my life expectancy is 85, now I'm at 5. I'm 100, let's say my life expectancy/duration is 2.

To implement I would bail on total bond and extend my duration/volatility out to 30 today, and reduce it to zero over the length of my retirement. So I take on a huge amount of rate risk today (the yield on a 5 year bond is similar to that of a 30 but the vol increases significantly at the far end of the curve) and gradually eliminating all yield above that of cash.

I like yield. Hell, if I live a very long time I might really need yield.

Also, what do I do with my stocks? Surely by far the largest source of portfolio volatility is my stocks, which I've still got even though I've eliminated rate risk in my bonds? Or do I eliminate equity risk as well going from 60-40-0 to 0-0-100 over time?

I will use myself as an example …

I need significant Cashflow for the next 6 years to meet my Retirement Goals. Once I start SS my Cashflow needs become minimal ( under 2%).

So duration matching is a very useful approach for positioning my assets to meet my Cashflow needs. Plus it’s not the entire portfolio so there are assets remaining for other Retirement or Estate Planning Goals.

You don’t have to duration match for your expected lifespan — I focus on a rolling 10 year period….

WoodSpinner

Sorry - are you matching duration or liabilities using a bond ladder? Your duration for this bucket would be about half the length of the ladder... I think.

[WoodSpinner]

I will readily admit to not understanding the rationale for duration matching at all - for individuals. Would welcome someone correcting my thinking!

I get it for an institution that may have a fixed payment at a certain future point and wants to squeeze every drop of yield out of their receivables before that date. But for an individual with an unknown "expiration date", I'm matching the duration of my bond portfolio to my life expectancy to eliminate rate risk as I get closer to death. That right?

Using me as an example - If I'm 55 my life expectancy is, say 85 so my duration should be a whopping 30 years? Versus the Agg's 6-7 years? So I massively extend my duration/volatility now and then each few years shorten it based on my reduced life expectancy? I'm eventually 65 and my life expectancy is say 82 so I reduce my duration from 30 to a still lengthy (and therefore volatile) 17?

And so on. At 80 say my life expectancy is 85, now I'm at 5. I'm 100, let's say my life expectancy/duration is 2.

To implement I would bail on total bond and extend my duration/volatility out to 30 today, and reduce it to zero over the length of my retirement. So I take on a huge amount of rate risk today (the yield on a 5 year bond is similar to that of a 30 but the vol increases significantly at the far end of the curve) and gradually eliminating all yield above that of cash.

I like yield. Hell, if I live a very long time I might really need yield.

Also, what do I do with my stocks? Surely by far the largest source of portfolio volatility is my stocks, which I've still got even though I've eliminated rate risk in my bonds? Or do I eliminate equity risk as well going from 60-40-0 to 0-0-100 over time?

I will use myself as an example …

I need significant Cashflow for the next 6 years to meet my Retirement Goals. Once I start SS my Cashflow needs become minimal ( under 2%).

So duration matching is a very useful approach for positioning my assets to meet my Cashflow needs. Plus it’s not the entire portfolio so there are assets remaining for other Retirement or Estate Planning Goals.

You don’t have to duration match for your expected lifespan — I focus on a rolling 10 year period….

WoodSpinner

Sorry - are you matching duration or liabilities using a bond ladder? Your duration for this bucket would be about half the length of the ladder... I think.

I have a choice in approach ….

I could use a Bond Fund with an approximate 3 year duration as you suggested, or purchase 6 bonds maturing in the appropriate year. Either way, I matching my Liabilities (expected Cashflow needs) to the duration of the bonds.

FWIW, I do both (individual bonds and a fund) to meet my needs.

WoodSpinner

[BrooklynInvest]

Using me as an example - If I'm 55 my life expectancy is, say 85 so my duration should be a whopping 30 years? Versus the Agg's 6-7 years? So I massively extend my duration/volatility now and then each few years shorten it based on my reduced life expectancy? I'm eventually 65 and my life expectancy is say 82 so I reduce my duration from 30 to a still lengthy (and therefore volatile) 17?

Don’t think so. This was covered recently. Your first duration is not 30 years but ~30/2. When you get to 65 years old, it’s dropped to 17/2.
viewtopic.php?p=7058849#p7058849

Perhaps. I'm a little hazy!

The duration of a LIABILITY-matching ladder would be a little less than half of the length of the ladder I get that.

But if I'm matching the DURATION of my investment horizon to the duration of my bond portfolio, which does eliminate rate risk at the expense of yield above cash over that period then surely it's my investment horizon? If I half it then I'm just going to cash earlier, no?

[Ben Mathew]

I guess I'll need some income from the bond-side too, so that would make duration matching suboptimal for me I guess?

No, duration matching does not reduce your income from the bond side. It just locks in current interest rates and makes you immune to changes in interest rates. Interest rates are equally likely to rise or fall, so you are not losing any expected return from locking in interest rates now. Not locking in interest rates now because you think they will go up would be a form of market timing.