Long Bonds in retirement - curiosity...

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
User avatar
vineviz
Posts: 10262
Joined: Tue May 15, 2018 1:55 pm

Re: Long Bonds in retirement - curiosity...

Post by vineviz »

abc132 wrote: Mon May 17, 2021 5:45 pm
vineviz wrote: Mon May 17, 2021 2:51 pm
abc132 wrote: Mon May 17, 2021 2:19 pm Think about whether having to start selling an 18 year duration asset in the next few years increases or reduces risk.
Hint: the correct answer is b) reduces risk.
If what you suggest was true, duration matching would always be detrimental.

If what I suggest is true, any plan that requires buying and selling long term bonds before their duration has to consider the desirable vs undesirable effect on the portfolio. How early this happens would be crucial to the best decision.
What I suggest IS true, though, a fact which does NOT imply that duration matching will "always be detrimental" but rather the opposite.

Duration matching using two or more bond funds will produce the same interest rate risk reduction as using an individual bond (or bonds) held to maturity. The only exceptions to that statement are minor, related to convexity and/or non-parallel shifts in the yield curve.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
User avatar
retired@50
Posts: 5925
Joined: Tue Oct 01, 2019 2:36 pm
Location: Living in the U.S.A.

Re: Long Bonds in retirement - curiosity...

Post by retired@50 »

daacrusher2001 wrote: Mon May 17, 2021 3:00 pm Perhaps just luck, but total bond has been a good investment for me over time. I wouldn't do a wholesale swap of total bond for LTT. I was mainly wondering if it's recommended to have a portion of my fixed income assets in long term treasuries and why.

I appreciate all the feedback from everyone.
By holding total bond, some of your fixed income assets ARE in long term treasuries.

From the Vanguard website: my emphasis.
This fund is designed to provide broad exposure to U.S. investment-grade bonds. Reflecting this goal, the fund invests in U.S. Treasuries and mortgage-backed securities of all maturities (short-, intermediate-, and long-term issues).
Regards,
This is one person's opinion. Nothing more.
User avatar
willthrill81
Posts: 25278
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: Long Bonds in retirement - curiosity...

Post by willthrill81 »

retired@50 wrote: Mon May 17, 2021 7:28 pm
daacrusher2001 wrote: Mon May 17, 2021 3:00 pm Perhaps just luck, but total bond has been a good investment for me over time. I wouldn't do a wholesale swap of total bond for LTT. I was mainly wondering if it's recommended to have a portion of my fixed income assets in long term treasuries and why.

I appreciate all the feedback from everyone.
By holding total bond, some of your fixed income assets ARE in long term treasuries.
That's true, but with an average effective duration of 8.5 years, it's definitely not a long-term bond fund. Also, it holds corporate bonds and not only Treasuries.

It's interesting that Vanguard had a long-term Treasuries fund long before it created a total bond market fund.
“Good and ill have not changed since yesteryear; nor are they one thing among Elves and Dwarves and another among Men.” J.R.R. Tolkien, The Lord of the Rings
User avatar
retired@50
Posts: 5925
Joined: Tue Oct 01, 2019 2:36 pm
Location: Living in the U.S.A.

Re: Long Bonds in retirement - curiosity...

Post by retired@50 »

willthrill81 wrote: Mon May 17, 2021 7:45 pm
retired@50 wrote: Mon May 17, 2021 7:28 pm
daacrusher2001 wrote: Mon May 17, 2021 3:00 pm Perhaps just luck, but total bond has been a good investment for me over time. I wouldn't do a wholesale swap of total bond for LTT. I was mainly wondering if it's recommended to have a portion of my fixed income assets in long term treasuries and why.

I appreciate all the feedback from everyone.
By holding total bond, some of your fixed income assets ARE in long term treasuries.
That's true, but with an average effective duration of 8.5 years, it's definitely not a long-term bond fund. Also, it holds corporate bonds and not only Treasuries.
For me, these facts ^^^ are a feature, not a bug.

Regards,
This is one person's opinion. Nothing more.
User avatar
vineviz
Posts: 10262
Joined: Tue May 15, 2018 1:55 pm

Re: Long Bonds in retirement - curiosity...

Post by vineviz »

retired@50 wrote: Mon May 17, 2021 7:52 pm
willthrill81 wrote: Mon May 17, 2021 7:45 pm
retired@50 wrote: Mon May 17, 2021 7:28 pm
daacrusher2001 wrote: Mon May 17, 2021 3:00 pm Perhaps just luck, but total bond has been a good investment for me over time. I wouldn't do a wholesale swap of total bond for LTT. I was mainly wondering if it's recommended to have a portion of my fixed income assets in long term treasuries and why.

I appreciate all the feedback from everyone.
By holding total bond, some of your fixed income assets ARE in long term treasuries.
That's true, but with an average effective duration of 8.5 years, it's definitely not a long-term bond fund. Also, it holds corporate bonds and not only Treasuries.
For me, these facts ^^^ are a feature, not a bug.
Perhaps, but they wouldn't be a feature to a long-term investor.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
User avatar
willthrill81
Posts: 25278
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: Long Bonds in retirement - curiosity...

Post by willthrill81 »

retired@50 wrote: Mon May 17, 2021 7:52 pm
willthrill81 wrote: Mon May 17, 2021 7:45 pm
retired@50 wrote: Mon May 17, 2021 7:28 pm
daacrusher2001 wrote: Mon May 17, 2021 3:00 pm Perhaps just luck, but total bond has been a good investment for me over time. I wouldn't do a wholesale swap of total bond for LTT. I was mainly wondering if it's recommended to have a portion of my fixed income assets in long term treasuries and why.

I appreciate all the feedback from everyone.
By holding total bond, some of your fixed income assets ARE in long term treasuries.
That's true, but with an average effective duration of 8.5 years, it's definitely not a long-term bond fund. Also, it holds corporate bonds and not only Treasuries.
For me, these facts ^^^ are a feature, not a bug.

Regards,
The fact that TBM covers all Treasuries (except TIPS) and investment-grade bonds is not a feature. Many have the mistaken belief that if a fund has the word 'total' in its name that it must be superior to funds without that word in their name.
“Good and ill have not changed since yesteryear; nor are they one thing among Elves and Dwarves and another among Men.” J.R.R. Tolkien, The Lord of the Rings
User avatar
retired@50
Posts: 5925
Joined: Tue Oct 01, 2019 2:36 pm
Location: Living in the U.S.A.

Re: Long Bonds in retirement - curiosity...

Post by retired@50 »

willthrill81 wrote: Mon May 17, 2021 8:08 pm
retired@50 wrote: Mon May 17, 2021 7:52 pm
willthrill81 wrote: Mon May 17, 2021 7:45 pm
retired@50 wrote: Mon May 17, 2021 7:28 pm
daacrusher2001 wrote: Mon May 17, 2021 3:00 pm Perhaps just luck, but total bond has been a good investment for me over time. I wouldn't do a wholesale swap of total bond for LTT. I was mainly wondering if it's recommended to have a portion of my fixed income assets in long term treasuries and why.

I appreciate all the feedback from everyone.
By holding total bond, some of your fixed income assets ARE in long term treasuries.
That's true, but with an average effective duration of 8.5 years, it's definitely not a long-term bond fund. Also, it holds corporate bonds and not only Treasuries.
For me, these facts ^^^ are a feature, not a bug.

Regards,
The fact that TBM covers all Treasuries (except TIPS) and investment-grade bonds is not a feature. Many have the mistaken belief that if a fund has the word 'total' in its name that it must be superior to funds without that word in their name.
willthrill81,
With all due respect, I realize that you're a knowledgeable, and frequent, poster here on Bogleheads.org. I've read and appreciate many of your posts.
But, I did qualify my statement with "For me". Does someone other than me get to decide what I consider a feature, really?

Regards,
This is one person's opinion. Nothing more.
User avatar
willthrill81
Posts: 25278
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: Long Bonds in retirement - curiosity...

Post by willthrill81 »

retired@50 wrote: Mon May 17, 2021 8:20 pm
willthrill81 wrote: Mon May 17, 2021 8:08 pm
retired@50 wrote: Mon May 17, 2021 7:52 pm
willthrill81 wrote: Mon May 17, 2021 7:45 pm
retired@50 wrote: Mon May 17, 2021 7:28 pm

By holding total bond, some of your fixed income assets ARE in long term treasuries.
That's true, but with an average effective duration of 8.5 years, it's definitely not a long-term bond fund. Also, it holds corporate bonds and not only Treasuries.
For me, these facts ^^^ are a feature, not a bug.

Regards,
The fact that TBM covers all Treasuries (except TIPS) and investment-grade bonds is not a feature. Many have the mistaken belief that if a fund has the word 'total' in its name that it must be superior to funds without that word in their name.
willthrill81,
With all due respect, I realize that you're a knowledgeable, and frequent, poster here on Bogleheads.org. I've read and appreciate many of your posts.
But, I did qualify my statement with "For me". Does someone other than me get to decide what I consider a feature, really?

Regards,
Fair enough. I missed that important detail and apologize.
“Good and ill have not changed since yesteryear; nor are they one thing among Elves and Dwarves and another among Men.” J.R.R. Tolkien, The Lord of the Rings
User avatar
grabiner
Advisory Board
Posts: 29936
Joined: Tue Feb 20, 2007 11:58 pm
Location: Columbia, MD

Re: Long Bonds in retirement - curiosity...

Post by grabiner »

If you are going to buy long-term bonds in retirement, it makes more sense to buy I-Bonds and long-term TIPS. If you buy a 30-year Treasury bond, you will get a guaranteed number of dollars in 2051, but you have no idea what those dollars will be worth; inflation 2.3% higher than expected over 30 years would cut the purchasing power in half. If you buy a 30-year TIPS (or buy an I-Bond and hold it for the full 30-years), you will get a guaranteed amount of purchasing power in 2051.
Wiki David Grabiner
anil686
Posts: 1225
Joined: Thu May 08, 2014 12:33 pm

Re: Long Bonds in retirement - curiosity...

Post by anil686 »

grabiner wrote: Mon May 17, 2021 8:47 pm If you are going to buy long-term bonds in retirement, it makes more sense to buy I-Bonds and long-term TIPS. If you buy a 30-year Treasury bond, you will get a guaranteed number of dollars in 2051, but you have no idea what those dollars will be worth; inflation 2.3% higher than expected over 30 years would cut the purchasing power in half. If you buy a 30-year TIPS (or buy an I-Bond and hold it for the full 30-years), you will get a guaranteed amount of purchasing power in 2051.
I don’t think I understand this well - just a question - is this the same for a LTT fund as well? For example, if rates rise significantly, I would expect them to fall in value but the coupons should make up for the drop in value correct? I guess it depends on how much the fund drops to figure out the break -even point....
User avatar
grabiner
Advisory Board
Posts: 29936
Joined: Tue Feb 20, 2007 11:58 pm
Location: Columbia, MD

Re: Long Bonds in retirement - curiosity...

Post by grabiner »

anil686 wrote: Mon May 17, 2021 9:02 pm
grabiner wrote: Mon May 17, 2021 8:47 pm If you are going to buy long-term bonds in retirement, it makes more sense to buy I-Bonds and long-term TIPS. If you buy a 30-year Treasury bond, you will get a guaranteed number of dollars in 2051, but you have no idea what those dollars will be worth; inflation 2.3% higher than expected over 30 years would cut the purchasing power in half. If you buy a 30-year TIPS (or buy an I-Bond and hold it for the full 30-years), you will get a guaranteed amount of purchasing power in 2051.
I don’t think I understand this well - just a question - is this the same for a LTT fund as well? For example, if rates rise significantly, I would expect them to fall in value but the coupons should make up for the drop in value correct? I guess it depends on how much the fund drops to figure out the break -even point....
Yes. The duration of Vanguard Long-Term Treasury is 18 years, even though the average maturity is 23 years, because some of the value of those Treasury bonds comes from coupons. Similarly, the duration of a 30-year zero-coupon Treasury is 30 years, but the duration of a 30-year coupon Treasury is about 23 years.

But you still face the same inflation risk. If Treasury yields rise by 1% next year, you will lose 18% of your fund value, and get back to the same number of dollars in 2039 as if rates hadn't changes. However, having the same number of dollars doesn't mean that you will have the same purchasing power; if inflation rises along with interest rates, the same dollars won't buy the same things.
Wiki David Grabiner
anil686
Posts: 1225
Joined: Thu May 08, 2014 12:33 pm

Re: Long Bonds in retirement - curiosity...

Post by anil686 »

grabiner wrote: Mon May 17, 2021 9:08 pm
anil686 wrote: Mon May 17, 2021 9:02 pm
grabiner wrote: Mon May 17, 2021 8:47 pm If you are going to buy long-term bonds in retirement, it makes more sense to buy I-Bonds and long-term TIPS. If you buy a 30-year Treasury bond, you will get a guaranteed number of dollars in 2051, but you have no idea what those dollars will be worth; inflation 2.3% higher than expected over 30 years would cut the purchasing power in half. If you buy a 30-year TIPS (or buy an I-Bond and hold it for the full 30-years), you will get a guaranteed amount of purchasing power in 2051.
I don’t think I understand this well - just a question - is this the same for a LTT fund as well? For example, if rates rise significantly, I would expect them to fall in value but the coupons should make up for the drop in value correct? I guess it depends on how much the fund drops to figure out the break -even point....
Yes. The duration of Vanguard Long-Term Treasury is 18 years, even though the average maturity is 23 years, because some of the value of those Treasury bonds comes from coupons. Similarly, the duration of a 30-year zero-coupon Treasury is 30 years, but the duration of a 30-year coupon Treasury is about 23 years.

But you still face the same inflation risk. If Treasury yields rise by 1% next year, you will lose 18% of your fund value, and get back to the same number of dollars in 2039 as if rates hadn't changes. However, having the same number of dollars doesn't mean that you will have the same purchasing power; if inflation rises along with interest rates, the same dollars won't buy the same things.

That is very helpful - thanks so much - makes sense now
abc132
Posts: 572
Joined: Thu Oct 18, 2018 1:11 am

Re: Long Bonds in retirement - curiosity...

Post by abc132 »

vineviz wrote: Mon May 17, 2021 6:56 pm Duration matching using two or more bond funds will produce the same interest rate risk reduction as using an individual bond (or bonds) held to maturity. The only exceptions to that statement are minor, related to convexity and/or non-parallel shifts in the yield curve.
I'm arguing non-parallel shifts matter for price when it comes to 30 year bonds and only holding for a year.

I recall one of your threads on how this really mattered for the interest rate your receive from long term bonds vs short term bonds. I feel I can argue the same for the price, which is what matters when you only hold for a year.

I would have trouble with someone that argued convexity matters for long vs short rates but not for long vs short price.

Is this your belief?
User avatar
vineviz
Posts: 10262
Joined: Tue May 15, 2018 1:55 pm

Re: Long Bonds in retirement - curiosity...

Post by vineviz »

abc132 wrote: Fri Jun 04, 2021 3:14 pm
vineviz wrote: Mon May 17, 2021 6:56 pm Duration matching using two or more bond funds will produce the same interest rate risk reduction as using an individual bond (or bonds) held to maturity. The only exceptions to that statement are minor, related to convexity and/or non-parallel shifts in the yield curve.
I'm arguing non-parallel shifts matter for price when it comes to 30 year bonds and only holding for a year.

I recall one of your threads on how this really mattered for the interest rate your receive from long term bonds vs short term bonds. I feel I can argue the same for the price, which is what matters when you only hold for a year.

I would have trouble with someone that argued convexity matters for long vs short rates but not for long vs short price.

Is this your belief?
I believe that no one with an investment horizon of 12 months should buy 30 year bonds.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
abc132
Posts: 572
Joined: Thu Oct 18, 2018 1:11 am

Re: Long Bonds in retirement - curiosity...

Post by abc132 »

vineviz wrote: Fri Jun 04, 2021 5:03 pm
abc132 wrote: Fri Jun 04, 2021 3:14 pm
vineviz wrote: Mon May 17, 2021 6:56 pm Duration matching using two or more bond funds will produce the same interest rate risk reduction as using an individual bond (or bonds) held to maturity. The only exceptions to that statement are minor, related to convexity and/or non-parallel shifts in the yield curve.
I'm arguing non-parallel shifts matter for price when it comes to 30 year bonds and only holding for a year.

I recall one of your threads on how this really mattered for the interest rate your receive from long term bonds vs short term bonds. I feel I can argue the same for the price, which is what matters when you only hold for a year.

I would have trouble with someone that argued convexity matters for long vs short rates but not for long vs short price.

Is this your belief?
I believe that no one with an investment horizon of 12 months should buy 30 year bonds.
Agreed, but the question is if someone with a 10 year expected horizon and 9 year current portfolio duration and some money to invest should
A) buy long term bonds to reach a 10 year portfolio duration and then be forced to sell in a year for duration matching when the portfolio now needs a 9 year duration
B) buy shorter term bond fund and let the portfolio match duration in one year

Selling a 30 year fund with a one year investment is exactly what would happen if one followed your prior recommendations that
1. duration matching of the portfolio is always optimal
2. only the duration of the entire portfolio matters

I'm suggesting that duration matching of the portfolio that requires you to buy and sell a 30 year fund within one year is sub-optimal. One would be better off in the above example just buying a shorter term (9-10 year) duration and letting the portfolio remain at a 9 year duration for one more year.

We can clearly see that duration matching of a portfolio is not always optimal for someone who is not currently matching duration - like someone considering long term bonds.

The reason should be obvious - that anything you buy should not need to be sold before it's duration unless you want to take on additional risks. Additional risks are taken if you purchase things you have to sell before their duration, which are mitigated to some extent by any benefits of a duration matched portfolio. This can all be done where you never sell a long term bond before it reaches duration but it requires a bit more planning and nuance than the absolute rules suggested by Vineviz.

1. duration matching of the portfolio is generally desirable
2. duration matching is optimal if it does not require you to sell things before you have owned them for their duration
3. if you have to sell before duration, it is optimal to balance the risk of owning a fund less than duration with the benefits of doing so

Case 3 would be for someone that finds themselves considering a higher duration but within the average duration of needing to rebalance downward.
User avatar
vineviz
Posts: 10262
Joined: Tue May 15, 2018 1:55 pm

Re: Long Bonds in retirement - curiosity...

Post by vineviz »

abc132 wrote: Fri Jun 04, 2021 6:52 pm
Agreed, but the question is if someone with a 10 year expected horizon and 9 year current portfolio duration and some money to invest should ....
Clearly the thing to do is to buy a 10 year bond and hold it to maturity.

Why make the portfolio more complicated for no reason?
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
User avatar
vineviz
Posts: 10262
Joined: Tue May 15, 2018 1:55 pm

Re: Long Bonds in retirement - curiosity...

Post by vineviz »

abc132 wrote: Fri Jun 04, 2021 6:52 pm I'm suggesting that duration matching of the portfolio that requires you to buy and sell a 30 year fund within one year is sub-optimal.
It's not suboptimal if it is keeping your bond duration matched to your investment horizon.

Buying a single individual bond and letting it "age" by a year produces the exact same result as rebalancing between two bond funds of fixed duration to reduce the average duration by a year.

Moving from 33% long-term bonds and 67% cash to 30% long-term bonds and 70% cash over the course of a year will produce the same benefits as letting a 10-year bond reduce naturally to a 9-year bond.

The former approach requires a little more effort, but otherwise the approaches are interchangeable.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
abc132
Posts: 572
Joined: Thu Oct 18, 2018 1:11 am

Re: Long Bonds in retirement - curiosity...

Post by abc132 »

vineviz wrote: Fri Jun 04, 2021 7:08 pm
abc132 wrote: Fri Jun 04, 2021 6:52 pm
Agreed, but the question is if someone with a 10 year expected horizon and 9 year current portfolio duration and some money to invest should ....
Clearly the thing to do is to buy a 10 year bond and hold it to maturity.

Why make the portfolio more complicated for no reason?

Because someone owning existing bonds and looking to buy some long term bonds with some additional investment may not have that as a cost-free option, and your claim is that all that matters is matching average duration - which can be done through addition without additional costs.

The choice to increase duration is
1) buy long term (say 18 year duration) bond to match duration and sell it prior to duration
2) deal with a suboptimal duration so that you are not buying and selling an 18 year duration fund in one year.

The specific example (new purchase and subsequent sale not matching duration with use) trumps the portfolio example of wanting duration to match use. Either one is suboptimal, and for the same reasons.

Increasing duration ---> good <--- Not buying and selling funds before duration

The optimal answer involves optimizing both of the above, for the singular reason that not matching use of a bond or bond fund with its duration is suboptimal. If you can't match duration with use without buying and selling a fund before duration, then something less than the ideal duration will be optimal.
User avatar
9-5 Suited
Posts: 671
Joined: Thu Jun 23, 2016 12:14 pm

Re: Long Bonds in retirement - curiosity...

Post by 9-5 Suited »

Bogleheads are very lucky to get a master class in bond investing from Vineviz in these threads. It’s some of the most valuable free advice around and it’s pretty amazing how often he is willing to go deep on the topic in a thread.

I had more or less “figured out” stock investing, but until recently I never had much need to learn about bonds and just threw a modest amount into TBM. Vineviz completely changed both my understanding of bond economics as well as my investment strategy. I have actively migrated to as close to a duration matched approach as I can within some practical limits based on fund availability. I no longer worry about the market price of my bond funds because I know losses from interest rate increases will be made up in due time by the higher future yields.

I strongly encourage those who dig deeper into the trench on the other side to take a quick moment to contemplate “what if I’m wrong and this guy is just right?”. I fought the impulse to agree with him for months, then once I saw it, it all clicked.

Thanks Vineviz!
User avatar
vineviz
Posts: 10262
Joined: Tue May 15, 2018 1:55 pm

Re: Long Bonds in retirement - curiosity...

Post by vineviz »

abc132 wrote: Fri Jun 04, 2021 8:51 pm Because someone owning existing bonds and looking to buy some long term bonds with some additional investment may not have that as a cost-free option, and your claim is that all that matters is matching average duration - which can be done through addition without additional costs.
It's not merely a "claim" that matching the duration of bonds to the investment horizon eliminates interest rate risk: it's a fact.

If someone prefers not to eliminate that risk due to alleged costs or complexity then I suppose that's their prerogative. But it's not as expensive nor as complex as you seem to think it is.
abc132 wrote: Fri Jun 04, 2021 8:51 pm If you can't match duration with use without buying and selling a fund before duration, then something less than the ideal duration will be optimal.
Nothing evil happens when you "buy and sell a fund before duration": if you keep the average duration of the portfolio aligned with the investment horizon, there's nothing to fear.

There's no use in continually repeating the word "suboptimal" without stopping to think about what that word means. You've got to specify what you're trying to optimize before you can claim that a strategy is somehow "suboptimal".

Matching bond duration to the investment horizon eliminates interest rate undertaken by the investor, and that is true regardless of the manner in which the matching is done.

If your argument is that some investors might prefer NOT to completely eliminate interest rate risk for some reason then I won't fight you on that claim. But the implication that there is some obvious problem with managing duration using bond funds is off base.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
abc132
Posts: 572
Joined: Thu Oct 18, 2018 1:11 am

Re: Long Bonds in retirement - curiosity...

Post by abc132 »

vineviz wrote: Fri Jun 04, 2021 10:38 pm
abc132 wrote: Fri Jun 04, 2021 8:51 pm Because someone owning existing bonds and looking to buy some long term bonds with some additional investment may not have that as a cost-free option, and your claim is that all that matters is matching average duration - which can be done through addition without additional costs.
It's not merely a "claim" that matching the duration of bonds to the investment horizon eliminates interest rate risk: it's a fact.


I certainly never argued that duration matching doesn't eliminate interest rate risk. Facts are better when they respond to the other person is saying. After typing in a long response, I think there are just too many fabrications on your part to have any real discussion. There are a few last responses from me so you can ponder how/why you failed to make your points to the other person you are conversing with:
vineviz wrote: Fri Jun 04, 2021 10:38 pm If someone prefers not to eliminate that risk due to alleged costs or complexity then I suppose that's their prerogative. But it's not as expensive nor as complex as you seem to think it is.
I don't think it's that expensive and definitely not complex. I think there is one optimum that includes costs and benefits of various actions. I certainly never said anything about complex or expensive so I'm curious what thought processes lead you to such thinking.
vineviz wrote: Fri Jun 04, 2021 7:08 pm Clearly the thing to do is to buy a 10 year bond and hold it to maturity.
Why make the portfolio more complicated for no reason?
I don't think it is complicated as you suggest above, but I would think that someone arguing that how we get to duration matching can matter should be able to use an example that can illustrate what they are trying to say.

[OT comments removed by admin LadyGeek]
vineviz wrote: Fri Jun 04, 2021 7:08 pm Matching bond duration to the investment horizon eliminates interest rate undertaken by the investor, and that is true regardless of the manner in which the matching is done.
Only if getting to an optimal portfolio is cost free and without risks such as buying and selling an 18 year fund in one year.

vineviz wrote: Fri Jun 04, 2021 7:08 pm If your argument is that some investors might prefer NOT to completely eliminate interest rate risk for some reason then I won't fight you on that claim. But the implication that there is some obvious problem with managing duration using bond funds is off base.


Nope that is not anything I have mentioned.
User avatar
vineviz
Posts: 10262
Joined: Tue May 15, 2018 1:55 pm

Re: Long Bonds in retirement - curiosity...

Post by vineviz »

abc132 wrote: Sat Jun 05, 2021 1:49 am
vineviz wrote: Fri Jun 04, 2021 7:08 pm Matching bond duration to the investment horizon eliminates interest rate undertaken by the investor, and that is true regardless of the manner in which the matching is done.
Only if getting to an optimal portfolio is cost free and without risks such as buying and selling an 18 year fund in one year.
There is no “only if” here. Matching bond duration to the investment horizon eliminates interest rate undertaken by the investor. Period.

If there are costs in maintaining a duration match they should be accounted for, obviously. In a world in which transaction costs are virtually zero (and not higher for bonds of long duration versus short) I want to see some example of what costs you think would make duration matching a problem.

And there is no risk associated directly with “ buying and selling an 18 year fund in one year”. Any “risk” would flow from a failure to maintain the match, not from the actions associated with maintaining it.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
User avatar
LadyGeek
Site Admin
Posts: 73137
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: Long Bonds in retirement - curiosity...

Post by LadyGeek »

I removed some off-topic comments. As a reminder, see: General Etiquette
At all times we must conduct ourselves in a respectful manner to other posters.
Wiki To some, the glass is half full. To others, the glass is half empty. To an engineer, it's twice the size it needs to be.
abc132
Posts: 572
Joined: Thu Oct 18, 2018 1:11 am

Re: Long Bonds in retirement - curiosity...

Post by abc132 »

vineviz wrote: Sat Jun 05, 2021 6:14 am
abc132 wrote: Sat Jun 05, 2021 1:49 am
vineviz wrote: Fri Jun 04, 2021 7:08 pm Matching bond duration to the investment horizon eliminates interest rate undertaken by the investor, and that is true regardless of the manner in which the matching is done.
Only if getting to an optimal portfolio is cost free and without risks such as buying and selling an 18 year fund in one year.
There is no “only if” here. Matching bond duration to the investment horizon eliminates interest rate undertaken by the investor. Period.

If there are costs in maintaining a duration match they should be accounted for, obviously. In a world in which transaction costs are virtually zero (and not higher for bonds of long duration versus short) I want to see some example of what costs you think would make duration matching a problem.

And there is no risk associated directly with “ buying and selling an 18 year fund in one year”. Any “risk” would flow from a failure to maintain the match, not from the actions associated with maintaining it.
I would love to have a discussion, but as I have said pre-moderation I do not think it is possible.
typical.investor
Posts: 2757
Joined: Mon Jun 11, 2018 3:17 am

Re: Long Bonds in retirement - curiosity...

Post by typical.investor »

abc132 wrote: Sun Jun 06, 2021 12:42 am
vineviz wrote: Sat Jun 05, 2021 6:14 am
abc132 wrote: Sat Jun 05, 2021 1:49 am
vineviz wrote: Fri Jun 04, 2021 7:08 pm Matching bond duration to the investment horizon eliminates interest rate undertaken by the investor, and that is true regardless of the manner in which the matching is done.
Only if getting to an optimal portfolio is cost free and without risks such as buying and selling an 18 year fund in one year.
There is no “only if” here. Matching bond duration to the investment horizon eliminates interest rate undertaken by the investor. Period.

If there are costs in maintaining a duration match they should be accounted for, obviously. In a world in which transaction costs are virtually zero (and not higher for bonds of long duration versus short) I want to see some example of what costs you think would make duration matching a problem.

And there is no risk associated directly with “ buying and selling an 18 year fund in one year”. Any “risk” would flow from a failure to maintain the match, not from the actions associated with maintaining it.
I would love to have a discussion, but as I have said pre-moderation I do not think it is possible.
Just so you two don't end up fighting with each other, I'll jump in and you can fight with me.

1) I agree there is some problem with trying to sell an 18 year duration fund to rebalance into a shorter duration fund to reduce overall duration. When you do you sell? It has to be 18 years before or you will be undertaking interest rate risk by selling before the fund has recovered. If there there is rate hike 25 year (for instance) before your duration mark, you'd have to wait until 7 years before your duration mark in order not to sell at a NAV loss. But if there (for instance) is another rate hike 8 years before your duration mark, you are going to have to sell at a NAV loss. I clearly see abc132 as being correct to raise this issue.

2) It probably won't matter that much if you are reducing equity allocation as retirement approaches because instead of having to sell that 18 year duration bond, you can adjust duration by adding a shorter duration fund out of cash

3) Using two bond funds, despite what vineviz claims, isn't going to completely do the trick. Even an Ultra-Short-Term Bond Fund will have a duration of say a year. So even if you match the 18 year fund and the 1 year fund to meet your duration of 5 years, when spending time comes, a 1% rise in rates could mean a 1% loss in the short term fund you plan to spend out of.

4) duration matching is the greatest thing since slice bread, but since duration is a function of yield, you may have to recalculate your duration each time the yield changes.

5) It surely easiest for a set date - say a house purchase in 5 years or 4 years of college starting in 10 years. For a 30 years retirement, your time horizon for your first year of spending and your last year of spending is quite different, and simply targeting a middle value is going to subject you to the problems illustrated in point 1. Using an 18 year fund and a 1 year fund that you start spending from is going to lead you in years 2,3,4,5 etc to have less money in the 1 year fund due to your spending, but require you to have more. So to match the shortening duration, you will have to sell the 18 year fund which could be problematic if there has been a rate hike as you don't have time to let the fund recover.

6) As for the claim that "Matching bond duration to the investment horizon eliminates interest rate undertaken by the investor. Period.", this is such an overstatement that no wonder there is strenous objection to the claim. The chosen phrasing renders it incorrect.

Here is what Morningstar I believe more accurately says:
Bond immunization is an investment strategy used to minimize the interest rate risk of bond investments by adjusting the portfolio duration to match the investor's investment time horizon.
Anyway, despite the complexity and limitations, I believe the underlying message is correct that matching bond duration to investment horizon can minimize interest rate risk.
abc132
Posts: 572
Joined: Thu Oct 18, 2018 1:11 am

Re: Long Bonds in retirement - curiosity...

Post by abc132 »

typical.investor wrote: Sun Jun 06, 2021 1:09 am
Just so you two don't end up fighting with each other, I'll jump in and you can fight with me.

1) I agree there is some problem with trying to sell an 18 year duration fund to rebalance into a shorter duration fund to reduce overall duration. When you do you sell? It has to be 18 years before or you will be undertaking interest rate risk by selling before the fund has recovered. If there there is rate hike 25 year (for instance) before your duration mark, you'd have to wait until 7 years before your duration mark in order not to sell at a NAV loss. But if there (for instance) is another rate hike 8 years before your duration mark, you are going to have to sell at a NAV loss. I clearly see abc132 as being correct to raise this issue.

2) It probably won't matter that much if you are reducing equity allocation as retirement approaches because instead of having to sell that 18 year duration bond, you can adjust duration by adding a shorter duration fund out of cash

3) Using two bond funds, despite what vineviz claims, isn't going to completely do the trick. Even an Ultra-Short-Term Bond Fund will have a duration of say a year. So even if you match the 18 year fund and the 1 year fund to meet your duration of 5 years, when spending time comes, a 1% rise in rates could mean a 1% loss in the short term fund you plan to spend out of.

4) duration matching is the greatest thing since slice bread, but since duration is a function of yield, you may have to recalculate your duration each time the yield changes.

5) It surely easiest for a set date - say a house purchase in 5 years or 4 years of college starting in 10 years. For a 30 years retirement, your time horizon for your first year of spending and your last year of spending is quite different, and simply targeting a middle value is going to subject you to the problems illustrated in point 1. Using an 18 year fund and a 1 year fund that you start spending from is going to lead you in years 2,3,4,5 etc to have less money in the 1 year fund due to your spending, but require you to have more. So to match the shortening duration, you will have to sell the 18 year fund which could be problematic if there has been a rate hike as you don't have time to let the fund recover.

6) As for the claim that "Matching bond duration to the investment horizon eliminates interest rate undertaken by the investor. Period.", this is such an overstatement that no wonder there is strenous objection to the claim. The chosen phrasing renders it incorrect.

Here is what Morningstar I believe more accurately says:
Bond immunization is an investment strategy used to minimize the interest rate risk of bond investments by adjusting the portfolio duration to match the investor's investment time horizon.
Anyway, despite the complexity and limitations, I believe the underlying message is correct that matching bond duration to investment horizon can minimize interest rate risk.
Vineviz has said the above don't really matter, you are reducing risk even if in the reality that plays out you lose money. Think of adding bonds to your portfolio and then watching bonds lose money while stocks gain. This would be a similar phenomenon to getting unlucky with interest rates, as improving expected value or portfolio characteristics does not guarantee an improvement in actual value or portfolio characteristics.


The core of my position is
1) The duration when we plan to spend the bonds may not be near the actual duration of our bonds
2) If we rebalance we must sell before duration
3) Our actual duration is not really a known value.
4) If we really need our bonds, it is likely to be in the shorter term because that is when sequence of risk shows up.

Consider:
1) rebalancing bonds to stocks according to a rule
2) rebalancing long bonds to take advantage of their volatility
3) scenarios where we need bonds earlier than planned

We should probably include 1-3 when we talk about a portfolio trying to take the least risk. When was the last time you did 1 and 2? (probably 2020, right?)

One 50% drop some time in the next 10 years for a 60/40 portfolio means 30% of the bonds should have something like a 5 year duration - just for one rebalancing event. That drops an expected 14 year use duration down to 11.3 (0.3*5 + 0.7*14 = 11.3), even if the sale of the bonds and repurchase are temporary. We take all that interest rate risk when rebalancing before our expected duration. If you rebalance, you minimize interest rate risk by using something less than the duration when you expect to spend the bonds and you can not eliminate interest rate risk because it is impossible to duration match your sale with rebalancing. Your AA, how and if you rebalance, these things matter.

For curiosities sake, I would ask anyone with a rebalancing rule for bonds to stocks and a rebalancing rule for long bonds to go through their portfolio for the past 25 years and calculate how much typical Boglehead rebalancing or actual lumpy spending reduces their actual bond duration as compared to when they expected to spend it permanently.

I believe the Boglehead intermediate bond fund is very sensible for many in their 60's when rebalancing, lumpy spending, and simplicity are included, although I have already said some long term bonds may be beneficial. It's something to think about, but the answer really depends on how you rebalance and how likely lumpy spending will be --> how sure you are you won't need to touch your bond fund.

I personally would want an actual portfolio analysis with expected actions before recommending long bonds for someone in their 60's. This would include their ability to tolerate short term losses of net value, their desire to rebalance, and their desire to continue rebalancing as they age. I like the intermediate recommendation over the duration matched recommendation as a general rule for the near or post retiree, and/or using something like I-bonds that just eliminates the interest rate risk automatically.

It seems likely to me that whatever your plan, in most realistic cases you will need to sell some bonds before they reach your planned duration. Adding long term bonds increases this risk because you have to sell some of them early if you duration match, and a more realistic scenario needs to counter the benefits of a supposedly perfectly duration matched portfolio - which is probably impossible in reality.

Anyone that rebalances and duration matches must sell bonds before their portfolio duration. You are always selling something now that was intended to be used at the portfolio duration and taking interest rate risk to reduce stock/bond AA risk (and rightly so!). The added interest rate risk of rebalancing when extending duration is countered by any benefits of extending duration.
Last edited by abc132 on Sun Jun 06, 2021 4:16 am, edited 14 times in total.
okwriter
Posts: 131
Joined: Mon Apr 12, 2021 2:00 pm

Re: Long Bonds in retirement - curiosity...

Post by okwriter »

typical.investor wrote: Sun Jun 06, 2021 1:09 am I agree there is some problem with trying to sell an 18 year duration fund to rebalance into a shorter duration fund to reduce overall duration. When you do you sell? It has to be 18 years before or you will be undertaking interest rate risk by selling before the fund has recovered.
I have a basic question about the above statement. Let's take a simple example with bonds and not bond funds. Consider 2 scenarios:

(1) You buy a 10-year duration bond. A year later, you'll have a 9-year bond.

(2) You buy two bonds: one with a 18-year duration and the other 2-year, so that your overall duration is 10 years. A year later, you sell both and buy a 9-year bond.

Assume that the yield curve only has parallel shifts. The wiki defines this as:
A parallel shift is when the long-term and short-term rates increase/decrease by the same amount.
Is the statement saying that (2) is more risky than (1)?

If there is a 1% increase in rates just before the end of year 1, the 18-(now 17)year bond's price will drop by 17%, 9-year by 9%, and 1-year by 1%. In scenario (2), you initially bought equal amounts of both the 18 and 2 year bonds, so your overall bond value drops by (17 + 1)/2 = 9%, which is the same as the 9-year bond.

So I don't see how (1) and (2) are any different.

I believe you can extend the above argument to bond funds as well.
typical.investor
Posts: 2757
Joined: Mon Jun 11, 2018 3:17 am

Re: Long Bonds in retirement - curiosity...

Post by typical.investor »

okwriter wrote: Sun Jun 06, 2021 3:38 am
typical.investor wrote: Sun Jun 06, 2021 1:09 am I agree there is some problem with trying to sell an 18 year duration fund to rebalance into a shorter duration fund to reduce overall duration. When you do you sell? It has to be 18 years before or you will be undertaking interest rate risk by selling before the fund has recovered.
I have a basic question about the above statement. Let's take a simple example with bonds and not bond funds. Consider 2 scenarios:

(1) You buy a 10-year duration bond. A year later, you'll have a 9-year bond.

(2) You buy two bonds: one with a 18-year duration and the other 2-year, so that your overall duration is 10 years. A year later, you sell both and buy a 9-year bond.

Assume that the yield curve only has parallel shifts. The wiki defines this as:
A parallel shift is when the long-term and short-term rates increase/decrease by the same amount.
Is the statement saying that (2) is more risky than (1)?

If there is a 1% increase in rates just before the end of year 1, the 18-(now 17)year bond's price will drop by 17%, 9-year by 9%, and 1-year by 1%. In scenario (2), you initially bought equal amounts of both the 18 and 2 year bonds, so your overall bond value drops by (17 + 1)/2 = 9%, which is the same as the 9-year bond.

So I don't see how (1) and (2) are any different.

I believe you can extend the above argument to bond funds as well.
I'm truly sorry for having to disagree as I'd like to sing "Kumbaya my Lord, kumbaya" and all that but no, I don't believe the above argument extends to bond funds. Just no. You bond math is very good by the way, please take that as a compliment and forgive me for shooting down your belief.

OK, I agree there isn't any difference between (1) and (2), so for a bond ladder - sure go for it.

For (1), a bond fund with a 10 year duration, a year later it will still have a 10 year duration. For (2), the 18 year/ 2-year combo will again still have an overall duration of 10 years. So, selling them both and buying a 9-year bond duration fund (or bond) in order to match your targeted investment duration, will mean that you have subject yourself to potential loss.

If rates climb 1% the day before you plan to sell in order to move your holdings to a 9-year duration, your 18 year fund might lose 18% and the 2 year fund 2% for an overall 10% loss. The 9 year you intend to buy will have lost 9%. OK, so you have 1% less money by selling your bond combo at a 10% loss and buying a replacement at a 9% loss.

If losing 1% of your money by using two bond funds through duration matching is your idea of under having eliminated interest rate risk, I don't know what to say. How it ultimately plays out is a little uncertain as it depends on both the yield curve and future rate movements.

Anyway, funds are managed to have a relatively constant duration. Individual bonds predictably reduce to zero and maturity approaches. You get the implications there right?
typical.investor
Posts: 2757
Joined: Mon Jun 11, 2018 3:17 am

Re: Long Bonds in retirement - curiosity...

Post by typical.investor »

abc132 wrote: Sun Jun 06, 2021 2:53 am
Vineviz has said the above don't really matter, you are reducing risk even if in the reality that plays out you lose money. Think of adding bonds to your portfolio and then watching bonds lose money while stocks gain. This would be a similar phenomenon to getting unlucky with interest rates, as improving expected value or portfolio characteristics does not guarantee an improvement in actual value or portfolio characteristics.
Yeah I know. But nobody can spend an expected value.

The simple fact is that duration pays a premium for taking risk. If you are going to use two bond funds to try to target your spending duration, you are adding the risk that your need to reduce duration will not be negatively impacted by changes in rates.

If you have a 1/10 chance of getting to a trillion and a 9/10 change of losing everything, the expected value looks pretty good. On average the payout is 100,000,000. But what if you are in the group where the risk shows up? Cat food? Homelessness?

Vineviz makes the same argument for factor investing. It's the same question though. What if your holding period is one of the long droughts and expected values don't materialize? Vineviz will claim so what, that is still the safe portfolio based on expected values even though in many cases the returns in the holding period could be negative.

So the question is how much risk is appropriate for you. If we could bank on expected value, then why isn't everyone 100% stocks? And if international stocks have higher expected returns, why aren't we all 100% international. I don't think that everyone should necessarily be 100% stocks and I don't think that everyone should necessarily target duration with two bond funds.

Those who build a long term individual TIPS ladder early might be on to something though. I do think TIPS have an insurance premium you have to pay though so would have to consider that. It might be small for the potential benefit. Again, it's an individual thing based on your preferences.
seajay
Posts: 146
Joined: Sat May 01, 2021 3:26 pm

Re: Long Bonds in retirement - curiosity...

Post by seajay »

typical.investor wrote: Sun Jun 06, 2021 4:41 am Those who build a long term individual TIPS ladder early might be on to something though. I do think TIPS have an insurance premium you have to pay though so would have to consider that. It might be small for the potential benefit. Again, it's an individual thing based on your preferences.
Is it not rather that other assets are riskier at providing certain objectives so involve a reward benefit on average for taking on that risk, but at-times/for-some may not fulfil the objective.

Approaching setting up a 20 year TIPS ladder to provide $10K/year of inflation adjusted income with a portfolio value of which $166K is being earmarked for that when TIPS real yields are +2%, later sees the portfolio value rise 46% to $243K and where TIPS real yields have transitioned to being -2% real. In both cases (back of napkin calculation) its the same outcome, the money buys a assured 10K/year of inflation adjusted income (maturing TIPS) for 20 years.

Lower yields are a reflection of higher prices/greater gains. Broadly washes, but where if anything the protection from purchasing TIPS is better when buying short/mid/long dated TIPS (ladder) when real yields are low/negative, as otherwise remaining in alternatives from relatively high valuation levels is more inclined to see low/poor subsequent rewards (or larger losses).

Whether a annuity might be employed instead is a coin-flip between longevity and leaving (or having access to) capital. Pass before a DIY annuity (TIPS ladder) is exhausted and that money is still within your estate. Survive to beyond the end of the ladder and the annuity collective pooling has that covered, continues paying.

There's a temptation to not employ a ladder when valuations are high however, at least from Trinity Study based valuations/metrics. Using the above figures for instance and $10K/year from $166K = 6% SWR. In contrast $10K/year from $243K = 4% SWR.
okwriter
Posts: 131
Joined: Mon Apr 12, 2021 2:00 pm

Re: Long Bonds in retirement - curiosity...

Post by okwriter »

typical.investor wrote: Sun Jun 06, 2021 4:16 am I don't believe the above argument extends to bond funds. Just no.
I'm glad we've settled that it's perfectly OK to sell 18-year duration bonds, before 18 years, for the purpose of duration matching, and there is no interest rate risk in doing so.

For bond funds, the argument may not extend perfectly, but if you're saying it doesn't extend at all, well, there's a full wiki article that disagrees. E.g.
BH Wiki wrote: We will discuss what happens to the fund or bond after a single interest-rate change. ... For instance, suppose a portfolio of bonds or funds is purchased with 5-year duration to meet a goal 5 years in the future, and is properly rebalanced to keep the duration equal to the investment horizon. With 4 years remaining, a parallel shift of the yield curve occurs. The investor would be indifferent to that interest rate change because of duration matching.

... The recommendation to "keep your duration equal to your need for the money" cannot be applied once and forgotten, but must be updated continuously (or more realistically, by rebalancing every 3-12 months into shorter-term funds). The only way to achieve duration matching automatically is to purchase a zero-coupon bond. Neither individual bonds nor funds will do this perfectly, although individual bonds come closer (particularly in low-interest rate environments).
For bond funds, note the recommendation above is 3-12 months.
typical.investor wrote: Sun Jun 06, 2021 4:16 am If rates climb 1% the day before you plan to sell in order to move your holdings to a 9-year duration, your 18 year fund might lose 18% and the 2 year fund 2% for an overall 10% loss. The 9 year you intend to buy will have lost 9%. OK, so you have 1% less money by selling your bond combo at a 10% loss and buying a replacement at a 9% loss.

If losing 1% of your money by using two bond funds through duration matching is your idea of under having eliminated interest rate risk, I don't know what to say. How it ultimately plays out is a little uncertain as it depends on both the yield curve and future rate movements.
What you are arguing about is a quantization issue. You're saying there's a potential loss of 1% per year if you duration-match once a year. Let's say, hypothetically, that you did it every day, so that you could gradually reduce your duration from 10 years to 9. Then there's no issue if interest rates rise. The wiki essentially says that 3-12 months is a good enough quantization of duration.

There's also another article (Bonds: Advanced topics) that says the same thing. I believe both articles cite books (and past BH forum posts).
For most purposes, shifting from intermediate- or longer-term bonds to shorter-term bonds as the need for capital approaches will cause little risk; also, in practice non-parallel shifts of the yield curve are not likely to cause very large changes in returns.
typical.investor
Posts: 2757
Joined: Mon Jun 11, 2018 3:17 am

Re: Long Bonds in retirement - curiosity...

Post by typical.investor »

okwriter wrote: Sun Jun 06, 2021 5:46 am
typical.investor wrote: Sun Jun 06, 2021 4:16 am I don't believe the above argument extends to bond funds. Just no.
I'm glad we've settled that it's perfectly OK to sell 18-year duration bonds, before 18 years, for the purpose of duration matching, and there is no interest rate risk in doing so.

For bond funds, the argument may not extend perfectly, but if you're saying it doesn't extend at all, well, there's a full wiki article that disagrees. E.g.
I don't see the BH Wiki disagreeing.
okwriter wrote: Sun Jun 06, 2021 5:46 am
BH Wiki wrote: We will discuss what happens to the fund or bond after a single interest-rate change. ... For instance, suppose a portfolio of bonds or funds is purchased with 5-year duration to meet a goal 5 years in the future, and is properly rebalanced to keep the duration equal to the investment horizon. With 4 years remaining, a parallel shift of the yield curve occurs. The investor would be indifferent to that interest rate change because of duration matching.
All that says is that having a duration of fours year in your bond holdings four years before your spending will make you indifferent because you had lowered your duration to four years.

It doesn't address the issue of what happens if rates had risen before you had lowered your duration. Note: that is where the risk is.
okwriter wrote: Sun Jun 06, 2021 5:46 am
BH Wiki wrote: ... The recommendation to "keep your duration equal to your need for the money" cannot be applied once and forgotten, but must be updated continuously (or more realistically, by rebalancing every 3-12 months into shorter-term funds). The only way to achieve duration matching automatically is to purchase a zero-coupon bond. Neither individual bonds nor funds will do this perfectly, although individual bonds come closer (particularly in low-interest rate environments).
For bond funds, note the recommendation above is 3-12 months.
Yeah, I agree that rebalancing every 3 months would have less risk than 12 months. If you believe the Wiki asserts that rebalancing every 3-12 months eliminates interest rate risk, I disagree. It states that continuously rebalancing will do the trick, but that 3-12 months is more realistic.

And again, this is for spending at one point in time. When you are talking about a 30-40 year retirement, you are going to be selling long bonds to shorten the duration and hit your 15 or 20 year target and they will have more potential to suffer NAV loss due to rate hikes.

I tried to look up the book they cite but didn't find it and can't say if the Michael Granito citation in Bond Portfolio Immunization (basis for the 3-12 month recommendation) was a) based on bond portfolios with fixed durations and applied to a long series of spending needs or b) based on individual bonds with one single spending needs. In other words, I remain unconvinced that the citation is applicable.
typical.investor wrote: Sun Jun 06, 2021 4:16 am If rates climb 1% the day before you plan to sell in order to move your holdings to a 9-year duration, your 18 year fund might lose 18% and the 2 year fund 2% for an overall 10% loss. The 9 year you intend to buy will have lost 9%. OK, so you have 1% less money by selling your bond combo at a 10% loss and buying a replacement at a 9% loss.

If losing 1% of your money by using two bond funds through duration matching is your idea of under having eliminated interest rate risk, I don't know what to say. How it ultimately plays out is a little uncertain as it depends on both the yield curve and future rate movements.
okwriter wrote: Sun Jun 06, 2021 5:46 am What you are arguing about is a quantization issue. You're saying there's a potential loss of 1% per year if you duration-match once a year. Let's say, hypothetically, that you did it every day, so that you could gradually reduce your duration from 10 years to 9. Then there's no issue if interest rates rise. The wiki essentially says that 3-12 months is a good enough quantization of duration.

There's also another article (Bonds: Advanced topics) that says the same thing. I believe both articles cite books (and past BH forum posts).
For most purposes, shifting from intermediate- or longer-term bonds to shorter-term bonds as the need for capital approaches will cause little risk; also, in practice non-parallel shifts of the yield curve are not likely to cause very large changes in returns.
If the Wiki is correct in its assertion that you can change from long term bonds to short term at any time, why doesn't everyone always hold long term bonds which yield more until they need to spend? I don't see any basis cited for that assertion and I reject it as unsupported.
User avatar
vineviz
Posts: 10262
Joined: Tue May 15, 2018 1:55 pm

Re: Long Bonds in retirement - curiosity...

Post by vineviz »

typical.investor wrote: Sun Jun 06, 2021 6:49 am It doesn't address the issue of what happens if rates had risen before you had lowered your duration. Note: that is where the risk is.
The risk here is only present to the extent that the investor lets the average duration of their bonds drift away from the actual investment time horizon. If annual rebalancing introduces a level of risk the investor is uncomfortable with they need only choose a more frequent rebalancing interval.
typical.investor wrote: Sun Jun 06, 2021 6:49 am And again, this is for spending at one point in time. When you are talking about a 30-40 year retirement, you are going to be selling long bonds to shorten the duration and hit your 15 or 20 year target and they will have more potential to suffer NAV loss due to rate hikes.
Whether the consumption is singular or spread out over time doesn't change anything. Over a long retirement, the time horizon shrinks very gradually which means the target duration is barely moving (i.e. it drops only by about 6 moths every year) and because the portfolio is presumably undergoing withdrawals the management of duration is much simpler. Just as if an investor would sell first from whichever asset class is overweight relative to their IPS, they would sell first from whichever bond fund is causing the most duration drift.

And in any case focusing on whether the portfolio duration has been stuck at 14 years while the investment time horizon has dwindled to 13.8 years is missing the forest for the trees. For one thing, that drift is purely in the control of the investor and occurs only if they allow it. For another thing, the most common alternative to matching duration is simply to choose a total bond market fund and forget it.

If people are freaked out that having a bond duration of 14 years is dangerous when the actual investment horizon is 13.8 years, it seems to me that they should be positively aghast that people are routine encouraged to buy a fund with a duration of 6 or 7 years under the same conditions.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
User avatar
vineviz
Posts: 10262
Joined: Tue May 15, 2018 1:55 pm

Re: Long Bonds in retirement - curiosity...

Post by vineviz »

abc132 wrote: Sun Jun 06, 2021 2:53 am
The core of my position is
1) The duration when we plan to spend the bonds may not be near the actual duration of our bonds
2) If we rebalance we must sell before duration
3) Our actual duration is not really a known value.
4) If we really need our bonds, it is likely to be in the shorter term because that is when sequence of risk shows up.
#1 relates to the ability of the investor to accurately estimate their investment time horizon to begin with. Most investors can do this reasonably well, especially if we're talking about retirement planning, but obviously there will always be uncertainty about the timing of future consumption. There are ways to incorporate that uncertainty into the planning and they are easy to implement. And once implemented the investor has an estimate of their time horizon they can use to proceed with duration matching.

#2 is not relevant to the discussion, since the investment time horizon is purely a function of consumption and rebalancing is not consumption.

#3 is the same as #1.

#4 also relates to #1, since that uncertainty merely folds into the estimate of investment time horizon. Once that estimate is generated, matching the bond duration to the horizon proceeds normally.

The consistent thread here seems to be a belief that uncertainty about the investment horizon makes duration matching impossible to do, but it doesn't seem to me to be the general case that investors have that much uncertainty. People routine form estimates about things like future income, longevity, retirement consumption, tax rates, etc. about which they are uncertain but it doesn't prevent us from doing any financial planning which involves those factors. Why should we consider uncertainty to a particularly acute problem only in this discussion?
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
User avatar
vineviz
Posts: 10262
Joined: Tue May 15, 2018 1:55 pm

Re: Long Bonds in retirement - curiosity...

Post by vineviz »

okwriter wrote: Sun Jun 06, 2021 3:38 am
typical.investor wrote: Sun Jun 06, 2021 1:09 am I agree there is some problem with trying to sell an 18 year duration fund to rebalance into a shorter duration fund to reduce overall duration. When you do you sell? It has to be 18 years before or you will be undertaking interest rate risk by selling before the fund has recovered.
I have a basic question about the above statement. Let's take a simple example with bonds and not bond funds. Consider 2 scenarios:

(1) You buy a 10-year duration bond. A year later, you'll have a 9-year bond.

(2) You buy two bonds: one with a 18-year duration and the other 2-year, so that your overall duration is 10 years. A year later, you sell both and buy a 9-year bond.

Assume that the yield curve only has parallel shifts. The wiki defines this as:
A parallel shift is when the long-term and short-term rates increase/decrease by the same amount.
Is the statement saying that (2) is more risky than (1)?

If there is a 1% increase in rates just before the end of year 1, the 18-(now 17)year bond's price will drop by 17%, 9-year by 9%, and 1-year by 1%. In scenario (2), you initially bought equal amounts of both the 18 and 2 year bonds, so your overall bond value drops by (17 + 1)/2 = 9%, which is the same as the 9-year bond.

So I don't see how (1) and (2) are any different.

I believe you can extend the above argument to bond funds as well.
Great summary, by the way.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
international001
Posts: 2017
Joined: Thu Feb 15, 2018 7:31 pm

Re: Long Bonds in retirement - curiosity...

Post by international001 »

vineviz wrote: Fri Jun 04, 2021 10:38 pm

It's not merely a "claim" that matching the duration of bonds to the investment horizon eliminates interest rate risk: it's a fact.
For the less bond-advanced.

We are talking here only about individual bonds, right? Inflaction aside, if you buy a 30 year bond, the risk is 0 because you'll get exactly what you sign up for (0 variance in the outcomes).

But if you buy a bond fund at year 0, (simplifying) every year you are buying that 29 year bond and buying a new 30 year bond, so there is risk any year in the future the value of the bond will be different based in all the interest rate changes. Am I wrong?

I also don't understand what is the big deal about bond risk. IF you have stocks in your portfolio, your total risk is going to be dominated by the risk of those stocks.
User avatar
vineviz
Posts: 10262
Joined: Tue May 15, 2018 1:55 pm

Re: Long Bonds in retirement - curiosity...

Post by vineviz »

international001 wrote: Sun Jun 06, 2021 7:04 pm But if you buy a bond fund at year 0, (simplifying) every year you are buying that 29 year bond and buying a new 30 year bond, so there is risk any year in the future the value of the bond will be different based in all the interest rate changes. Am I wrong?
If you want to manage interest rate risk using bond funds you simply use two or more bond funds to keep the average duration on a continuously declining path, mimicking the natural behavior of the individual bond.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
okwriter
Posts: 131
Joined: Mon Apr 12, 2021 2:00 pm

Re: Long Bonds in retirement - curiosity...

Post by okwriter »

vineviz wrote: Sun Jun 06, 2021 7:41 am
okwriter wrote: Sun Jun 06, 2021 3:38 am
typical.investor wrote: Sun Jun 06, 2021 1:09 am I agree there is some problem with trying to sell an 18 year duration fund to rebalance into a shorter duration fund to reduce overall duration. When you do you sell? It has to be 18 years before or you will be undertaking interest rate risk by selling before the fund has recovered.
I have a basic question about the above statement. Let's take a simple example with bonds and not bond funds. Consider 2 scenarios:

(1) You buy a 10-year duration bond. A year later, you'll have a 9-year bond.

(2) You buy two bonds: one with a 18-year duration and the other 2-year, so that your overall duration is 10 years. A year later, you sell both and buy a 9-year bond.

Assume that the yield curve only has parallel shifts. The wiki defines this as:
A parallel shift is when the long-term and short-term rates increase/decrease by the same amount.
Is the statement saying that (2) is more risky than (1)?

If there is a 1% increase in rates just before the end of year 1, the 18-(now 17)year bond's price will drop by 17%, 9-year by 9%, and 1-year by 1%. In scenario (2), you initially bought equal amounts of both the 18 and 2 year bonds, so your overall bond value drops by (17 + 1)/2 = 9%, which is the same as the 9-year bond.

So I don't see how (1) and (2) are any different.

I believe you can extend the above argument to bond funds as well.
Great summary, by the way.
Thanks. Just an attempt to simplify things, to get at how duration-matching works.
international001
Posts: 2017
Joined: Thu Feb 15, 2018 7:31 pm

Re: Long Bonds in retirement - curiosity...

Post by international001 »

vineviz wrote: Sun Jun 06, 2021 7:18 pm
international001 wrote: Sun Jun 06, 2021 7:04 pm But if you buy a bond fund at year 0, (simplifying) every year you are buying that 29 year bond and buying a new 30 year bond, so there is risk any year in the future the value of the bond will be different based in all the interest rate changes. Am I wrong?
If you want to manage interest rate risk using bond funds you simply use two or more bond funds to keep the average duration on a continuously declining path, mimicking the natural behavior of the individual bond.
Yes.. I understand that.

I guess I don't understand the big hype about buying an individual bond (or a set of mutual funds, as you say) to make sure risk is absolute 0. Over 30 years, a combination of stocks and a bond funds (let's say of permanent intermediate duration) has a very low chance of underperforming a 30 year bond.
User avatar
vineviz
Posts: 10262
Joined: Tue May 15, 2018 1:55 pm

Re: Long Bonds in retirement - curiosity...

Post by vineviz »

international001 wrote: Mon Jun 07, 2021 3:58 am I guess I don't understand the big hype about buying an individual bond (or a set of mutual funds, as you say) to make sure risk is absolute 0. Over 30 years, a combination of stocks and a bond funds (let's say of permanent intermediate duration) has a very low chance of underperforming a 30 year bond.
I don't know that there is much "hype" going on here, just an attempt to explain/understand how an investor can effectively manage risk that they want to manage.

Clearly investors with equity-heavy portfolios face less interest rate risk than investors with bond-heavy portfolios. Most investors are risk-averse (or, more specifically, loss-averse) which is what leads them to own bonds to begin with. There's no rational reason that such investors would knowingly or willingly accept an uncompensated risk (e.g. interest rate risk) when it is easily avoided.

I agree that this is not the most burning problem that most investors need to tackle, but it is a problem which is easily and cheaply solved. And when the solution results in having a higher expected return and less risk, it's probably worth spending a few words discussing IMHO.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
Escapevelocity
Posts: 606
Joined: Mon Feb 18, 2019 8:32 am

Re: Long Bonds in retirement - curiosity...

Post by Escapevelocity »

vineviz wrote: Sun May 16, 2021 9:19 pm
Explorer wrote: Sun May 16, 2021 8:35 pm
Beensabu wrote: Sun May 16, 2021 8:25 pm
Explorer wrote: Sun May 16, 2021 7:51 pm
vineviz wrote: Sun May 16, 2021 10:03 am

This actually has the effect of INCREASING the interest rate risk of the portfolio, which is the opposite of what most investors think they are achieving.
I am not buying your assertion.. unless you explain in simple terms.
It's reinvestment risk. You're betting that short-term rates will go up enough soon enough (and stay that way) that rolling over short-term bonds will have you better off than accepting the long-term rate that you know will give you $X at Y date.
Fair. But, shifting to short bond does NOT increase interest rate risk.
It does, because of what dbr just explained. When you own short-term bonds but have a long-term investment goal, interest rate risk takes the form of reinvestment risk: if yields fall you'll be forced to reinvest the interest in lower yielding bonds, causing your total return to be less than expected. Getting less-than-expected returns as a result of a change in interest rates is the very definition of interest rate risk.
I agree with the premise, but here is where my self-professed market timing attitude on rates comes into play. With bonds, I refuse to accede to the nobody knows nothing motto.
abc132
Posts: 572
Joined: Thu Oct 18, 2018 1:11 am

Re: Long Bonds in retirement - curiosity...

Post by abc132 »

typical.investor wrote: Sun Jun 06, 2021 4:41 am
abc132 wrote: Sun Jun 06, 2021 2:53 am
Vineviz has said the above don't really matter, you are reducing risk even if in the reality that plays out you lose money. Think of adding bonds to your portfolio and then watching bonds lose money while stocks gain. This would be a similar phenomenon to getting unlucky with interest rates, as improving expected value or portfolio characteristics does not guarantee an improvement in actual value or portfolio characteristics.
Yeah I know. But nobody can spend an expected value.

The simple fact is that duration pays a premium for taking risk. If you are going to use two bond funds to try to target your spending duration, you are adding the risk that your need to reduce duration will not be negatively impacted by changes in rates.

If you have a 1/10 chance of getting to a trillion and a 9/10 change of losing everything, the expected value looks pretty good. On average the payout is 100,000,000. But what if you are in the group where the risk shows up? Cat food? Homelessness?

Vineviz makes the same argument for factor investing. It's the same question though. What if your holding period is one of the long droughts and expected values don't materialize? Vineviz will claim so what, that is still the safe portfolio based on expected values even though in many cases the returns in the holding period could be negative.

So the question is how much risk is appropriate for you. If we could bank on expected value, then why isn't everyone 100% stocks? And if international stocks have higher expected returns, why aren't we all 100% international. I don't think that everyone should necessarily be 100% stocks and I don't think that everyone should necessarily target duration with two bond funds.

Those who build a long term individual TIPS ladder early might be on to something though. I do think TIPS have an insurance premium you have to pay though so would have to consider that. It might be small for the potential benefit. Again, it's an individual thing based on your preferences.
I thought of using the 10% worst case to match bond duration, similar to how we plan for a safe withdrawal rate instead of using expected values. My guess is that poor sequence of events shorten bond duration, but because there can be several causes of a poor sequence it is something I would need to look further into.
abc132
Posts: 572
Joined: Thu Oct 18, 2018 1:11 am

Re: Long Bonds in retirement - curiosity...

Post by abc132 »

vineviz wrote: Sun Jun 06, 2021 7:37 am
abc132 wrote: Sun Jun 06, 2021 2:53 am
The core of my position is
1) The duration when we plan to spend the bonds may not be near the actual duration of our bonds
2) If we rebalance we must sell before duration
3) Our actual duration is not really a known value.
4) If we really need our bonds, it is likely to be in the shorter term because that is when sequence of risk shows up.
#1 relates to the ability of the investor to accurately estimate their investment time horizon to begin with. Most investors can do this reasonably well, especially if we're talking about retirement planning, but obviously there will always be uncertainty about the timing of future consumption. There are ways to incorporate that uncertainty into the planning and they are easy to implement. And once implemented the investor has an estimate of their time horizon they can use to proceed with duration matching.

I'm not convinced we can plan how we plan to spend bonds in the future very accurately. In my portfolio bonds are the thing that must respond to stock risk. You have shown the models how small changes in AA can make or break a portfolio over time. That permanent spending can happen early, later, or never. It's not about handling uncertainty, it is about amount of uncertainty in portfolio growth that can really change when bonds need to be spent. It depends how our portfolio grows/falls and how we end needing to spend, and whether we plan for average or lower probability scenarios (SWR, 10% case, etc).

vineviz wrote: Sun Jun 06, 2021 7:37 am #2 is not relevant to the discussion, since the investment time horizon is purely a function of consumption and rebalancing is not consumptiuon.
It is relevant if you argue people can routinely calculate an accurate duration, despite the evidence of the contrary.

It is relevant to my discussion, as it is my belief.

Responding to it would move the discussion forward, and I would appreciate your feedback.

vineviz wrote: Sun Jun 06, 2021 7:37 am
The consistent thread here seems to be a belief that uncertainty about the investment horizon makes duration matching impossible to do, but it doesn't seem to me to be the general case that investors have that much uncertainty. People routine form estimates about things like future income, longevity, retirement consumption, tax rates, etc. about which they are uncertain but it doesn't prevent us from doing any financial planning which involves those factors. Why should we consider uncertainty to a particularly acute problem only in this discussion?
That's certainly not what I said at all, nor do I think I align with your beliefs about most others.

Respectfully, I'm not sure how to correct such misrepresentations.

I do prefer to express my own beliefs over having them done for me.
NoRegret
Posts: 308
Joined: Sat Dec 16, 2017 2:00 am
Location: California

Re: Long Bonds in retirement - curiosity...

Post by NoRegret »

typical.investor wrote: Sat May 15, 2021 8:28 pm
daacrusher2001 wrote: Sat May 15, 2021 7:30 pm I'm wondering if I should consider using part of the money allocated to Total Bond to something like Vanguard Long Term Treasury. If I'm lucky, I'll get 30 years in retirement...hoping for at least 20.
You can but know that the duration of the Vanguard Long Term Treasury is about 18 years. That means it will recover loss from a rate hike today in 18 years. So yeah it seems to fit your time line.

But also know that this is true tomorrow and 20 and 30 years from today as well. So if there is a rate hike in 10 years, you will need 18 years from that point to recover with an expected 10 - 20 years (confident you are going to make it to the longish side) in your horizon remaining.

Of course, 30 years is a pretty long time and I suspect there will be a typical recession sometime where rates drop to stimulate the economy and long bonds in particular do well. If you are holding long bonds and well into your retirement, that might be a good time to rebalance out of any you are holding.

Or go with intermediate and not worry about the timing.
I looked into it a bit a while ago, not sure if it's right, but here it goes. Assume rates increase linearly for X years and then stay constant, it will take (X/2 + duration) years to get even, up to X = twice duration. This is assuming one owns a bond fund. So if we round-trip past four decades of rate reduction it will take 36 yrs to get back to even.

Somebody please check my math.
Market timer targeting long term cycles -- aiming for several key decisions per asset class per decade
User avatar
vineviz
Posts: 10262
Joined: Tue May 15, 2018 1:55 pm

Re: Long Bonds in retirement - curiosity...

Post by vineviz »

abc132 wrote: Mon Jun 07, 2021 3:28 pm I'm not convinced we can plan how we plan to spend bonds in the future very accurately.
The way this sentence is worded makes me think there's a misunderstanding which underlies your argument.

Estimating the investment horizon doesn't involve planning how to "spend bonds" per se, but instead involves estimating future portfolio withdrawals for the purpose of consumption. It's this overall spending that matters, not specifically the bond spending. They're related, clearly, since it's generally optimal to withdraw proportionately from stocks and bonds to fund consumption. But asset allocation and investment horizon are technically distinct.

Additionally, the "very accurately" requirement is a red herring. For one thing the requirement is both arbitrary and subjective. For another the vast majority of people can estimate their investment horizon "accurately enough" to know whether their investment horizon is short-, medium- or long-term.
abc132 wrote: Mon Jun 07, 2021 3:28 pm
vineviz wrote: Sun Jun 06, 2021 7:37 am #2 is not relevant to the discussion, since the investment time horizon is purely a function of consumption and rebalancing is not consumptiuon.
It is relevant if you argue people can routinely calculate an accurate duration, despite the evidence of the contrary.
It's irrelevant because rebalancing has nothing to do with estimating the investor's time horizon OR calculating their portfolio's duration.

abc132 wrote: Mon Jun 07, 2021 3:28 pm
vineviz wrote: Sun Jun 06, 2021 7:37 am
The consistent thread here seems to be a belief that uncertainty about the investment horizon makes duration matching impossible to do, but it doesn't seem to me to be the general case that investors have that much uncertainty. People routine form estimates about things like future income, longevity, retirement consumption, tax rates, etc. about which they are uncertain but it doesn't prevent us from doing any financial planning which involves those factors. Why should we consider uncertainty to a particularly acute problem only in this discussion?
That's certainly not what I said at all, nor do I think I align with your beliefs about most others.
You just wrote this: "I'm not convinced we can plan how we plan to spend bonds in the future very accurately." I don't know how to interpret that statement except as an expression of a "belief that uncertainty about the investment horizon makes duration matching impossible ".

If you have a different belief then I invite you to clarify what you meant.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
User avatar
vineviz
Posts: 10262
Joined: Tue May 15, 2018 1:55 pm

Re: Long Bonds in retirement - curiosity...

Post by vineviz »

Escapevelocity wrote: Mon Jun 07, 2021 10:47 am I agree with the premise, but here is where my self-professed market timing attitude on rates comes into play. With bonds, I refuse to accede to the nobody knows nothing motto.
The reason that market timing doesn't work isn't that "nobody knows nothing". It's that you don't know anything useful about the future that isn't already incorporated into prices by the market.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
international001
Posts: 2017
Joined: Thu Feb 15, 2018 7:31 pm

Re: Long Bonds in retirement - curiosity...

Post by international001 »

vineviz wrote: Mon Jun 07, 2021 9:37 am
Clearly investors with equity-heavy portfolios face less interest rate risk than investors with bond-heavy portfolios. Most investors are risk-averse (or, more specifically, loss-averse) which is what leads them to own bonds to begin with. There's no rational reason that such investors would knowingly or willingly accept an uncompensated risk (e.g. interest rate risk) when it is easily avoided.
Let me try to make my question another way. Let's say your AA is 60% stocks and 40% bonds. You have 30 years planned to live, an you plan to keep that AA.

Is it worth planning to get a 30 year bond to have 0 risk on the bond part?
I'd like to minimize the risk on my overall portfolio. Isn't the bond part an insignificant part of it?
grok87
Posts: 9624
Joined: Tue Feb 27, 2007 9:00 pm

Re: Long Bonds in retirement - curiosity...

Post by grok87 »

international001 wrote: Mon Jun 07, 2021 6:54 pm
vineviz wrote: Mon Jun 07, 2021 9:37 am
Clearly investors with equity-heavy portfolios face less interest rate risk than investors with bond-heavy portfolios. Most investors are risk-averse (or, more specifically, loss-averse) which is what leads them to own bonds to begin with. There's no rational reason that such investors would knowingly or willingly accept an uncompensated risk (e.g. interest rate risk) when it is easily avoided.
Let me try to make my question another way. Let's say your AA is 60% stocks and 40% bonds. You have 30 years planned to live, an you plan to keep that AA.

Is it worth planning to get a 30 year bond to have 0 risk on the bond part?
I'd like to minimize the risk on my overall portfolio. Isn't the bond part an insignificant part of it?
if by 30 year bond you mean 30 year tips, then i would say yes. most people's retirement spending needs are inflation-sensitive, not nominal
RIP Mr. Bogle.
User avatar
vineviz
Posts: 10262
Joined: Tue May 15, 2018 1:55 pm

Re: Long Bonds in retirement - curiosity...

Post by vineviz »

international001 wrote: Mon Jun 07, 2021 6:54 pm Let me try to make my question another way. Let's say your AA is 60% stocks and 40% bonds. You have 30 years planned to live, an you plan to keep that AA.

Is it worth planning to get a 30 year bond to have 0 risk on the bond part?
I'd like to minimize the risk on my overall portfolio. Isn't the bond part an insignificant part of it?
You wouldn't want a single 30-year bond in this case, but if you DID buy a single 30-year bond as the "40%" in your 60/40 portfolio then the bond would account for roughly half the overall volatility of the portfolio. I wouldn't call that insignificant.

And if you truly wanted to minimize the risk on the overall portfolio then a ladder of TIPS spanning the full 30 years would be the way to do that.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
abc132
Posts: 572
Joined: Thu Oct 18, 2018 1:11 am

Re: Long Bonds in retirement - curiosity...

Post by abc132 »

vineviz wrote: Mon Jun 07, 2021 4:29 pm
abc132 wrote: Mon Jun 07, 2021 3:28 pm I'm not convinced we can plan how we plan to spend bonds in the future very accurately.
The way this sentence is worded makes me think there's a misunderstanding which underlies your argument.

Estimating the investment horizon doesn't involve planning how to "spend bonds" per se, but instead involves estimating future portfolio withdrawals for the purpose of consumption. It's this overall spending that matters, not specifically the bond spending. They're related, clearly, since it's generally optimal to withdraw proportionately from stocks and bonds to fund consumption. But asset allocation and investment horizon are technically distinct.

Additionally, the "very accurately" requirement is a red herring. For one thing the requirement is both arbitrary and subjective. For another the vast majority of people can estimate their investment horizon "accurately enough" to know whether their investment horizon is short-, medium- or long-term.
I plan to continue to change AA based on probability of outcomes - actually using probability to inform decisions. I've settled on 70/30 right into early retirement, but where to go from there depends on portfolio value relative to expenses. You can check out your own thread that picked AA based on minimizing retirement outcome spread as an example of why portfolio value can matter for AA selection. "Withdrawing in proportion to what AA? " would be my response to what you suggest - that is not known in advance for anyone that uses probability modelling and has a diverse set of potential outcomes.

Since 2009, we have gotten about 15% annual nominal stock returns. People that modelled a likely SWR can spend much more than what they thought they could spend, and the AA they thought they would need for a particular set of objectives is nowhere near the AA they now need for those same objectives. They may even have changed their objectives (self-sufficiency to legacy/charity), and those actually using probability in a meaningful way would generally not expect to know their AA in advance. I want significant bonds if I am focused on self sufficiency, and very few bonds if I am focused on legacy and/or charity. The difference in time horizon of those two things is 50 years.

vineviz wrote: Sun Jun 06, 2021 7:37 am It's irrelevant because rebalancing has nothing to do with estimating the investor's time horizon OR calculating their portfolio's duration.
I was hoping you would think about providing something that illustrates your statement, or provides some support for it.
vineviz wrote: Sun Jun 06, 2021 7:37 am You just wrote this: "I'm not convinced we can plan how we plan to spend bonds in the future very accurately." I don't know how to interpret that statement except as an expression of a "belief that uncertainty about the investment horizon makes duration matching impossible ".

If you have a different belief then I invite you to clarify what you meant.
I think many Bogleheads have conservative enough plans that they may find their "survival portfolio" with 2% SWR and lots of bonds quickly turns into a legacy portfolio without the need for any bonds. Those who have done probabilistic modelling can see how quickly and easily this happens. Or you can just look at the last 12 years of history --> I've seen a lot of posts of people that surely thought they wanted bonds, but have now dumped them for other vehicles. The data I see says many people are unlikely to know what they want very far into the future, whether it is AA, spend rate, presence of bonds, etc.
vineviz wrote: Mon Jun 07, 2021 4:29 pm The consistent thread here seems to be a belief that uncertainty about the investment horizon makes duration matching impossible to do, but it doesn't seem to me to be the general case that investors have that much uncertainty.
How could something I've said may be worthwhile and gave an example of be considered impossible by me? Can it be beneficial, sure, I've said that, and even gave an example, although the chosen duration to match differed.

Of course we can try to predict. Duration matching is literally the effort to do so, is an inexact matching as our future expenses are not perfectly known, and by definition of being a process to attempt to match duration this process can not be impossible. Is calculating a SWR impossible? Of course not, but the question becomes how actionable/useful a value can be. In probability analysis, the more important value is often not the average or expected value but the dispersion of outcomes. If you replace the word "impossible" which has a really high bar to be true with something like "is duration matching required for a good enough portfolio", then the answer is still likely no.

As to investor behavior, I would say to read a few posts about people dumping their bonds all together, switching their AA, changing from preservation to legacy, or finding out they can have 3x the expenditures they thought 10 years ago they would have today. I don't see much evidence that investors can predict anything about the future very well, and that is because the future is quite a bit unknown. Investors can have trouble predicting even their own behavior. The uncertainty in the future and in self makes the least parts of optimization easily possible but largely a waste of time.

Even for someone managing interest rate risk, there are alternatives to extending duration. As one example, I bonds can be used to help manage interest rate risk with much less duration. In decision making you learn which decisions really effect the outcome, and learn to focus on those. Someone choosing mid term bond over longer duration funds is simply not one of those decisions that materially affect the outcome of a reasonable portfolio, nor is duration or bonds the only way to manage interest rate risk. People may put commodities, gold, or alternatives into their portfolio. Nobody should be aghast at a "good enough" portfolio, at least not on a site dedicated to a Boglehead philosophy.
Post Reply