Should we discourage the bond tent?

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Topic Author
scout1
Posts: 37
Joined: Thu Oct 18, 2018 3:26 pm

Should we discourage the bond tent?

Post by scout1 » Mon Oct 29, 2018 3:09 pm

I have seen a number of discussions on BH referencing the Bond Tent as well as a number of BH members saying that they are going to use the strategy. To me it looks like a bad strategy and I believe BH'ers should push back on it.

The bond tent violates the most fundamental rule of asset allocation: money should be allocated to match time horizon. If you’re 65 and your portfolio needs to last 30 years, over-weighting bonds doesn’t make sense and when you’re 85 with a 10 year horizon, over-weighting equity to compensate for being underweight earlier is even worse. The strategy locks you into lower risk when you have a higher risk tolerance and higher risk when you have a lower risk tolerance. No hypothetical numbers are even needed, this is self-evidently true.

The appeal of the strategy seems to be emotional since people get to overweight bonds when they have a large portfolio. This violates the BH philosophy of avoiding emotional decision making. The "size effect" that Kitces references I believe is a bit of a fallacy. Clearly, if the stock market crashes when you experience a large unexpected withdrawal it is going to be bad, but the bond tent just shifts that risk from the beginning of retirement and arguable makes it even worse if it happens near the end of your retirement (you can't time the market or unexpected withdrawals) while also guaranteeing an asset-liability mismatch.


Below is an explanation of the Bond Tent.
https://www.kitces.com/blog/managing-po ... -red-zone/

BigJohn
Posts: 1704
Joined: Wed Apr 02, 2014 11:27 pm

Re: Should we discourage the bond tent?

Post by BigJohn » Tue Oct 30, 2018 1:10 am

Full disclosure, I'm using something similar to a bond tent and don't think is driven by emotions or violates any fundamental principles. Kitces comes at the reason for higher bonds early in retirement based on sequence of return risk. I agree but will come at it another way that gets to the same answer.

I have no pension and don't plan to start SS until age 70 so all my living expenses for the first 10 years of retirement have to come from my portfolio. I set my bond allocation at the start of retirement based on the concept of a liability matching portfolio. I put ~30 years of minimal living expense in safe assets (high quality, intermediate term bonds). Doing this increased my bond allocation by +15-20% from where it had been for several years. Now, fast forward 20 years or so with only ~10 years of life expectancy left. The liability I need to match is far lower so the $$ on the safe asset side are far lower. Hopefully during those 20 years the risk side of my portfolio (TSM/TISM) has done well and is now worth more than when I started. So my bond allocation has fallen and.... a bond tent was created.

I agree the strategy isn't a one size fits all for everyone but for many it's appropriate. Several comments from just my perspective to specific points you made.
scout1 wrote:
Mon Oct 29, 2018 3:09 pm
If you’re 65 and your portfolio needs to last 30 years, over-weighting bonds doesn’t make sense
I disagree, sequence of return matters and can matter a lot during the first 10 years of retirement.
scout1 wrote:
Mon Oct 29, 2018 3:09 pm
and when you’re 85 with a 10 year horizon, over-weighting equity to compensate for being underweight earlier is even worse.
Higher equity allocation later isn't compensating for anything. It's an outcome of higher ability to take risk because you have the ultimate "investment time horizon" in your sites.
scout1 wrote:
Mon Oct 29, 2018 3:09 pm
The strategy locks you into lower risk when you have a higher risk tolerance and higher risk when you have a lower risk tolerance. No hypothetical numbers are even needed, this is self-evidently true.
Based on above, I feel I have lower risk tolerance now when my money has to last 30 years and will have higher risk tolerance in 20 years when my money only has to last 10 years.
scout1 wrote:
Mon Oct 29, 2018 3:09 pm
Clearly, if the stock market crashes when you experience a large unexpected withdrawal it is going to be bad,
What you may be missing is that it doesn't take a large unexpected withdrawal to trigger a sequence of return problem during a market correction. When 100% of your living expenses are coming from some part of your portfolio, having a higher bond allocation helps mitigate that risk. If I had pensions, SS, etc to cover basic living expenses the answer might be different.

Again, I agree that the concept is not appropriate for everyone but as Kitces points out, for many it can be a prudent decision.

visualguy
Posts: 1276
Joined: Thu Jan 30, 2014 1:32 am

Re: Should we discourage the bond tent?

Post by visualguy » Tue Oct 30, 2018 1:40 am

BigJohn wrote:
Tue Oct 30, 2018 1:10 am
Based on above, I feel I have lower risk tolerance now when my money has to last 30 years and will have higher risk tolerance in 20 years when my money only has to last 10 years.
Hard to say... Those last years can be very expensive. Expenses related to health, long-term care, moving to a CCRC. Some of this can hit one spouse first, and then the other spouse later, which magnifies the cost even further. I've seen all this happen in my family, and I would definitely avoid taking risks with the money in my later years.

In particular, I think folks who can't (or don't want to) rely on family to take care of them need to be particularly careful with this because throwing money at it is pretty much all you can do for help at that point. Sometimes it doesn't even take more than a bad fall for an old person, and the money spigots have to be opened. Something even more serious, and the spigots have to be opened wider. You need the money to be there and not significantly evaporated by some downturn or financial crisis...

User avatar
nisiprius
Advisory Board
Posts: 38253
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: Should we discourage the bond tent?

Post by nisiprius » Tue Oct 30, 2018 6:28 am

1) I don't think we should "encourage" or "discourage" the use of anything unless there's really clear or compelling evidence. Pretty much nothing about retirement savings glide paths is clear or compelling. The data since 1926 only gives us three non-overlapping thirty-year periods to look at, and people did not really start trying to use individual portfolio withdrawals until the rise of 401(k) defined-contribution plans.

We should discourage reliance on any "safe withdrawal rates" or glide path studies, as anything beyond a rough rule of thumb or a ballpark planning number.

2) Kitces' sample "bond tent" involves 30% stocks at age 65. If we run the Vanguard "Nest Egg Calculator," a Monte Carlo simulator, accepting 30 years as the time frame (as Kitces does), setting a withdrawal rate of 4%, and the allocation to 30/70, it shows a 90% chance of the portfolio lasting 30 years:

Image

If we change the allocation to 70/30, the tool gives us exactly the same portfolio survival probability:

Image

In my personal opinion, this is evidence that at 4% withdrawal rate is way too high, but it is not evidence for any issue with a 30/70 allocation.

The finding--that varying allocations within the general range of perhaps 25/75 to 75/25 has very little effect on portfolio survival--is quite robust and is true over many different simulation tools. The only situations in which high stock allocations produce much higher survival rates are those in which you insist on an aggressive withdrawal rate and accept low portfolio survival rates. (For example, with a 6% withdrawal rate, Vanguard's tool shows 37% survival for 30/70, 62% for 70/30... and 66% for 100% stocks. But how many people can truly accept a 66% portfolio survival probability?)

There is not, and has never been any evidence for there being much difference in portfolio success rates at conservative withdrawal percentage rates, and for any allocation that isn't extreme. There are reasons "discourage" the use of either 100% stocks or 100% bonds at age 65, but there's nothing demonstrably terrible with 30/70, even without a rising stock allocation.

3) All of the analyses of glide paths don't even try to deal in any good way with the issues of what "risk" really means in retirement, and factually what the risk tolerance of retirees really is. Almost all of them just sort of set up a 5% failure rate as acceptable, without seriously investing whether that is or is not acceptable, how big the changes in life style are if it turns out that the 5% chance is actually going to happen, and what the impact of those changes are.

We should not assume that retirees do not know their own risk tolerance, and we should not force them (or at least strongly noodge them) to take on more risk for their own good.

What we can do, and what Kitces himself has done, is to emphasize that the Kitces and Pfau proposals--which are not IMHO supported by anything more than fairly cursory backtesting--should be framed in terms of lowering stock allocation for ages close to retirement, rather than inaccurately framing them as proposals for retirees in later retirement to increase their stock allocation. The proposal, originally frames as a "rising glide path," only rises because it starts from a lower starting point.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.

User avatar
Johnsson
Posts: 223
Joined: Mon Jul 17, 2017 2:28 pm

Re: Should we discourage the bond tent?

Post by Johnsson » Tue Oct 30, 2018 11:59 am

BigJohn wrote:
Tue Oct 30, 2018 1:10 am
I have no pension and don't plan to start SS until age 70 so all my living expenses for the first 10 years of retirement have to come from my portfolio. I set my bond allocation at the start of retirement based on the concept of a liability matching portfolio. I put ~30 years of minimal living expense in safe assets (high quality, intermediate term bonds). Doing this increased my bond allocation by +15-20% from where it had been for several years. Now, fast forward 20 years or so with only ~10 years of life expectancy left. The liability I need to match is far lower so the $$ on the safe asset side are far lower. Hopefully during those 20 years the risk side of my portfolio (TSM/TISM) has done well and is now worth more than when I started. So my bond allocation has fallen and.... a bond tent was created.
This.

An ever increasing percentage of our population will be dependent upon only savings plus SS in retirement as pensions are slowly phased out. Waiting until 70 to start collecting SS is prudent, as it provides the largest 'guaranteed' ( :confused ) continuous income stream most of will have available (assuming an SPIA isn't purchased). This leaves us with two different funding requirements... from retirement until age 70 (before SS) and after 70 with SS.

The 'bond tent' (or CD ladder as much of ours will be) allows a true minimization of risk during the pre-SS years (yes, there are still risks r/t inflation and rising rates). Once SS hits the level of draw will be greatly reduced allowing a much higher level of risk to be assumed.

Our draw should be about 3 to 3.7% pre-SS and 0 to 1.1% with SS. Our allocation should start ~50/50 and finish (at 70) at ~80/20, though I'll probably tone it back to 70/30 or 60/40.

Without the bond tent we would face a much higher level of risk in those early (pre-SS years)

Assuming a reasonable level of spending a bond tent is what will virtually guarantee our successful retirement.

BigJohn
Posts: 1704
Joined: Wed Apr 02, 2014 11:27 pm

Re: Should we discourage the bond tent?

Post by BigJohn » Tue Oct 30, 2018 6:26 pm

visualguy wrote:
Tue Oct 30, 2018 1:40 am
Hard to say... Those last years can be very expensive. Expenses related to health, long-term care, moving to a CCRC. Some of this can hit one spouse first, and then the other spouse later, which magnifies the cost even further. I've seen all this happen in my family, and I would definitely avoid taking risks with the money in my later years.

In particular, I think folks who can't (or don't want to) rely on family to take care of them need to be particularly careful with this because throwing money at it is pretty much all you can do for help at that point. Sometimes it doesn't even take more than a bad fall for an old person, and the money spigots have to be opened. Something even more serious, and the spigots have to be opened wider. You need the money to be there and not significantly evaporated by some downturn or financial crisis...
Fair points as I’ve seen it happen with my parents and my wife. While the increasing stock allocation is sort of the default based on spending from the liability matching side, it isn’t inevitable. Twenty years is a long time and depending on how things evolve (my health, the health care system, increases in life expectancy, etc), I can always adjust and keep my stock allocation more modest.

TN_Boy
Posts: 834
Joined: Sat Jan 17, 2009 12:51 pm

Re: Should we discourage the bond tent?

Post by TN_Boy » Tue Oct 30, 2018 6:39 pm

scout1 wrote:
Mon Oct 29, 2018 3:09 pm
I have seen a number of discussions on BH referencing the Bond Tent as well as a number of BH members saying that they are going to use the strategy. To me it looks like a bad strategy and I believe BH'ers should push back on it.

The bond tent violates the most fundamental rule of asset allocation: money should be allocated to match time horizon. If you’re 65 and your portfolio needs to last 30 years, over-weighting bonds doesn’t make sense and when you’re 85 with a 10 year horizon, over-weighting equity to compensate for being underweight earlier is even worse. The strategy locks you into lower risk when you have a higher risk tolerance and higher risk when you have a lower risk tolerance. No hypothetical numbers are even needed, this is self-evidently true.

The appeal of the strategy seems to be emotional since people get to overweight bonds when they have a large portfolio. This violates the BH philosophy of avoiding emotional decision making. The "size effect" that Kitces references I believe is a bit of a fallacy. Clearly, if the stock market crashes when you experience a large unexpected withdrawal it is going to be bad, but the bond tent just shifts that risk from the beginning of retirement and arguable makes it even worse if it happens near the end of your retirement (you can't time the market or unexpected withdrawals) while also guaranteeing an asset-liability mismatch.


Below is an explanation of the Bond Tent.
https://www.kitces.com/blog/managing-po ... -red-zone/
A number of BHer's don't like the "bond tent," so it is far from a universally supported idea.

Personally, I rather like the concept, because I think a major stock market crash early in retirement really is one of the biggest potential problems I'm likely to have.

If equities do great in early retirement, those with a decent chunk of bonds will still do fine. And if equities do horribly, making withdrawals from bonds rather than beaten-up stocks will feel like a victory for the strategy. And, every year that goes by gets closer to taking SS and taking more pressure off the portfolio.

An 85 year old may or may not have a 10 year horizon. I know a number of people who were like, dead, at 85. They didn't need any money at all. And even if they were alive, you need a lot less money to handle 10 years than 30.

It's important; to be poor at 85 when your retirement finances were reasonable, not one, but two things had to happen.

1) you had awful market returns (or exceptional expenses) and
2) you have to be alive.

As Nisprius points out, studies haven't shone anything magic about a particular stock allocation in retirement. I like the idea of taking some risk off the table to reduce sequence of risk issues early.

columbia
Posts: 1555
Joined: Tue Aug 27, 2013 5:30 am

Re: Should we discourage the bond tent?

Post by columbia » Tue Oct 30, 2018 6:40 pm

I suppose an extreme version would be to sell all of your equities 5 years before retirement and slowly dollar cost average back in to a 30%-50% equity level.

That might be a terrible idea, but sure would cut out most of the nastiest sequence of returns risks.

flyingaway
Posts: 2252
Joined: Fri Jan 17, 2014 10:19 am

Re: Should we discourage the bond tent?

Post by flyingaway » Tue Oct 30, 2018 7:02 pm

When you have won the game, why keep playing?
When you have a large portfolio, you need to protect it first.

User avatar
peetsperk
Posts: 172
Joined: Wed Feb 04, 2015 4:02 pm

Re: Should we discourage the bond tent?

Post by peetsperk » Tue Oct 30, 2018 8:49 pm

nisiprius wrote:
Tue Oct 30, 2018 6:28 am
1) I don't think we should "encourage" or "discourage" the use of anything unless there's really clear or compelling evidence. Pretty much nothing about retirement savings glide paths is clear or compelling. The data since 1926 only gives us three non-overlapping thirty-year periods to look at, and people did not really start trying to use individual portfolio withdrawals until the rise of 401(k) defined-contribution plans.

We should discourage reliance on any "safe withdrawal rates" or glide path studies, as anything beyond a rough rule of thumb or a ballpark planning number.

2) Kitces' sample "bond tent" involves 30% stocks at age 65. If we run the Vanguard "Nest Egg Calculator," a Monte Carlo simulator, accepting 30 years as the time frame (as Kitces does), setting a withdrawal rate of 4%, and the allocation to 30/70, it shows a 90% chance of the portfolio lasting 30 years:

Image

If we change the allocation to 70/30, the tool gives us exactly the same portfolio survival probability:

Image

In my personal opinion, this is evidence that at 4% withdrawal rate is way too high, but it is not evidence for any issue with a 30/70 allocation.

The finding--that varying allocations within the general range of perhaps 25/75 to 75/25 has very little effect on portfolio survival--is quite robust and is true over many different simulation tools. The only situations in which high stock allocations produce much higher survival rates are those in which you insist on an aggressive withdrawal rate and accept low portfolio survival rates. (For example, with a 6% withdrawal rate, Vanguard's tool shows 37% survival for 30/70, 62% for 70/30... and 66% for 100% stocks. But how many people can truly accept a 66% portfolio survival probability?)

There is not, and has never been any evidence for there being much difference in portfolio success rates at conservative withdrawal percentage rates, and for any allocation that isn't extreme. There are reasons "discourage" the use of either 100% stocks or 100% bonds at age 65, but there's nothing demonstrably terrible with 30/70, even without a rising stock allocation.

3) All of the analyses of glide paths don't even try to deal in any good way with the issues of what "risk" really means in retirement, and factually what the risk tolerance of retirees really is. Almost all of them just sort of set up a 5% failure rate as acceptable, without seriously investing whether that is or is not acceptable, how big the changes in life style are if it turns out that the 5% chance is actually going to happen, and what the impact of those changes are.

We should not assume that retirees do not know their own risk tolerance, and we should not force them (or at least strongly noodge them) to take on more risk for their own good.

What we can do, and what Kitces himself has done, is to emphasize that the Kitces and Pfau proposals--which are not IMHO supported by anything more than fairly cursory backtesting--should be framed in terms of lowering stock allocation for ages close to retirement, rather than inaccurately framing them as proposals for retirees in later retirement to increase their stock allocation. The proposal, originally frames as a "rising glide path," only rises because it starts from a lower starting point.
Thanks nisiprius! As usual, you bring clarity to a theoretical question for which there is not a definitive answer. I appreciate the time you took to walk us through a summary of the finer points. Like other Bogleheads, whenever I seek greater understanding on this website, I look for the butterfly.
:sharebeer

OffGridder
Posts: 42
Joined: Thu Jul 23, 2015 8:03 am
Location: Eastern WA.

Re: Should we discourage the bond tent?

Post by OffGridder » Tue Oct 30, 2018 9:22 pm

nisiprius wrote:
Tue Oct 30, 2018 6:28 am

The finding--that varying allocations within the general range of perhaps 25/75 to 75/25 has very little effect on portfolio survival--is quite robust and is true over many different simulation tools. The only situations in which high stock allocations produce much higher survival rates are those in which you insist on an aggressive withdrawal rate and accept low portfolio survival rates. (For example, with a 6% withdrawal rate, Vanguard's tool shows 37% survival for 30/70, 62% for 70/30... and 66% for 100% stocks. But how many people can truly accept a 66% portfolio survival probability?)

There is not, and has never been any evidence for there being much difference in portfolio success rates at conservative withdrawal percentage rates, and for any allocation that isn't extreme. There are reasons "discourage" the use of either 100% stocks or 100% bonds at age 65, but there's nothing demonstrably terrible with 30/70, even without a rising stock allocation.

The above could serve as encouragement to maintain a constant conservative asset allocation (low allocation to equities) during retirement. The possible justification for having a more aggressive allocation to equities would be if you are looking to leave a large bequeath. The retirement decumulation phase is not the time to be swinging for the fences, particularly when there is no need and doing so will not improve your portfolio survival.
"Goodness is the only investment that never fails." | H.D. Thoreau

NoHeat
Posts: 308
Joined: Sun Sep 18, 2016 10:13 am

Re: Should we discourage the bond tent?

Post by NoHeat » Tue Oct 30, 2018 9:30 pm

The “portfolio size effect” is not explained in the Kitces blog article cited in the OP, even though Kitces claims it to be the problem that the bond tent is intended to fix.

For your chances of dying broke, it’s not obvious to me why a dip in the stock market is inherently more devastating at age 65 than at 75, which seems to be the claim of this supposed “size effect.”

I can think of another problem the bond tent actually can fix — a delay in building a floor with SS would be a good reason to temporarily hold more bonds, until SS begins. Because after SS begins you don’t need to withdraw as much from the portfolio, and therefore a stock dip won’t hurt as much, so it’s ok to hold more stocks then.

But delaying SS doesn’t seem to be what Kitces had in mind with his supposed “size effect.” So I’m still at a loss to figure out what Kitces is claiming.

stlutz
Posts: 5093
Joined: Fri Jan 02, 2009 1:08 am

Re: Should we discourage the bond tent?

Post by stlutz » Tue Oct 30, 2018 9:43 pm

Suppose somebody retires at age 65 and they decide to annuitize their 401K. Between this and Social Security, their living needs are covered. They also have a Roth IRA and some taxable savings that are invested mostly in stocks. These assets could be used for various desires or extraordinary needs, or they may be passed onto heirs.

Over the course of their retirement, the net present value of the annuity declines. The value of their equity investments most likely rises. Are they violating the most fundamental rule of asset allocation or have they constructed a sensible plan?

Now, I changed the original question slightly by replacing bonds with an annuity, but the same principle would apply with a bond portfolio. If I want to be able to draw a constant dollar amount our of my bond portfolio each year, I can do this as long as my portfolio duration matches my investing horizon--i.e. if I'm planning to live 30 years then my duration should be 15 (since, on average, I'll spend the money in 15 years). Over time the value of my bond portfolio declines. My equity portfolio might not.

It seems to me that the objection is not so much to the rising glidepath result but to the way of framing the issue.

BigJohn
Posts: 1704
Joined: Wed Apr 02, 2014 11:27 pm

Re: Should we discourage the bond tent?

Post by BigJohn » Tue Oct 30, 2018 10:35 pm

NoHeat wrote:
Tue Oct 30, 2018 9:30 pm
For your chances of dying broke, it’s not obvious to me why a dip in the stock market is inherently more devastating at age 65 than at 75, which seems to be the claim of this supposed “size effect.”
Not sure what size effect has to do with it but here's an explanation of why a correction is worse early in retirement.
https://www.kitces.com/blog/understandi ... d-decades/

User avatar
willthrill81
Posts: 10122
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: Should we discourage the bond tent?

Post by willthrill81 » Tue Oct 30, 2018 10:51 pm

BigJohn wrote:
Tue Oct 30, 2018 10:35 pm
NoHeat wrote:
Tue Oct 30, 2018 9:30 pm
For your chances of dying broke, it’s not obvious to me why a dip in the stock market is inherently more devastating at age 65 than at 75, which seems to be the claim of this supposed “size effect.”
Not sure what size effect has to do with it but here's an explanation of why a correction is worse early in retirement.
https://www.kitces.com/blog/understandi ... d-decades/
Indeed. From that post:
As the chart above reveals, there is a strong and consistent relationship between safe withdrawal rates for a 60/40 portfolio and the real returns of equities during the first decade of retirement. Extremely low real returns (e.g., below 0%) are highly damaging to the sustainable spending level (and in fact any compounded real returns below about 3% are still in the 4% to 5.5% safe withdrawal rate range). On the other end, there is a positive sequence of return effect visible as well; with a good first decade (e.g., 12%+ returns), the worst safe withdrawal rates were still at least 7% and often higher! In other words, just as a bad first decade can be so severe with ongoing withdrawals that a subsequent market rebound just isn’t enough to recover, but a good first decade can be so positive than even a subsequent bear market can’t ruin the outcome!
The bottom line is that for fixed withdrawals, in the first decade, the portfolio is very likely to either suffer so greatly that eventual failure is inevitable or to grow to such a point that even a nasty bear market doesn't threaten its survival. There isn't much middle ground. But that's for fixed withdrawals. To the extent that withdrawals are flexible, sequence of returns risk goes down. For strictly flexible withdrawals (i.e. fixed percentage of portfolio withdrawals), there is literally no sequence of returns risk because the effect of returns is 'conveyed' directly to the withdrawals.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

AlohaJoe
Posts: 4493
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: Should we discourage the bond tent?

Post by AlohaJoe » Tue Oct 30, 2018 11:06 pm

BigJohn wrote:
Tue Oct 30, 2018 10:35 pm
NoHeat wrote:
Tue Oct 30, 2018 9:30 pm
For your chances of dying broke, it’s not obvious to me why a dip in the stock market is inherently more devastating at age 65 than at 75, which seems to be the claim of this supposed “size effect.”
Not sure what size effect has to do with it but here's an explanation of why a correction is worse early in retirement.
https://www.kitces.com/blog/understandi ... d-decades/
The article doesn't show that at all!

If anything, the article shows that corrections in early retirement are meaningless to retirement success and people should stop worrying about them.

Kitces shows that 1 year nominal returns have no real relationship to SWR. (0.2 correlation, I think.)

Even 10 year nominal returns (which is stretching the definition of "early retirement" given that 10 years is over half of retirement for most people) don't have a meaningful correlation -- 0.44. A number so low most people would call it meaningless.

It is clear that nominal returns, which is what a correction is, don't affect SWRs.

What does affect SWRs is inflation, which is why Kitces charts of real returns have a strong correlation. But there's also a bit meaningless...I mean to say that real returns over the first 10 years predict retirement success is.....well, other than real returns what else is even involved in the model? You're just begging the question.

b0B
Posts: 513
Joined: Fri Oct 19, 2018 11:39 am

Re: Should we discourage the bond tent?

Post by b0B » Tue Oct 30, 2018 11:57 pm

The “Rising Equity Glidepath” is a very silly idea. The main silliness is robotically following a predetermined glide path regardless of whatever circumstances arise. (And yes, most people wouldn't do that, but computers testing models do exactly that.)

But "Bond Tent" seems a much more nebulous term, so it's harder to pin down for faults.

It's the kind of fuzzy term where two reasonable people could have a conversation:

"Oh I see what you're doing. You're doing a Bond Tent"
"No I'm not. I'm not doing a Bond Tent at all."
"Oh yes you are.
"Oh no I'm not."
"Oh yes you are.
"Oh no I'm not."
etc...

User avatar
HomerJ
Posts: 12842
Joined: Fri Jun 06, 2008 12:50 pm

Re: Should we discourage the bond tent?

Post by HomerJ » Wed Oct 31, 2018 12:02 am

scout1 wrote:
Mon Oct 29, 2018 3:09 pm
The bond tent violates the most fundamental rule of asset allocation: money should be allocated to match time horizon. If you’re 65 and your portfolio needs to last 30 years, over-weighting bonds doesn’t make sense and when you’re 85 with a 10 year horizon, over-weighting equity to compensate for being underweight earlier is even worse. The strategy locks you into lower risk when you have a higher risk tolerance and higher risk when you have a lower risk tolerance. No hypothetical numbers are even needed, this is self-evidently true.
Any time someone tells me something is "self-evidently true", I immediately doubt what they are saying. :)

I don't think you are taking sequence of return risks into account.

Risk while withdrawing money is very different from the risk when accumulating money. I think what you wrote above is incorrect.

65 year-olds have a LOWER risk tolerance than an 85-year old.

Sure, some of their money is considered "long-term", but a large percentage of it is going to be withdrawn in the short-term, and that shouldn't be risked in the stock market.
The J stands for Jay

randomguy
Posts: 7676
Joined: Wed Sep 17, 2014 9:00 am

Re: Should we discourage the bond tent?

Post by randomguy » Wed Oct 31, 2018 12:48 am

nisiprius wrote:
Tue Oct 30, 2018 6:28 am
1) I don't think we should "encourage" or "discourage" the use of anything unless there's really clear or compelling evidence. Pretty much nothing about retirement savings glide paths is clear or compelling. The data since 1926 only gives us three non-overlapping thirty-year periods to look at, and people did not really start trying to use individual portfolio withdrawals until the rise of 401(k) defined-contribution plans.

We should discourage reliance on any "safe withdrawal rates" or glide path studies, as anything beyond a rough rule of thumb or a ballpark planning number.

2) Kitces' sample "bond tent" involves 30% stocks at age 65. If we run the Vanguard "Nest Egg Calculator," a Monte Carlo simulator, accepting 30 years as the time frame (as Kitces does), setting a withdrawal rate of 4%, and the allocation to 30/70, it shows a 90% chance of the portfolio lasting 30 years:

Image

If we change the allocation to 70/30, the tool gives us exactly the same portfolio survival probability:

Image

In my personal opinion, this is evidence that at 4% withdrawal rate is way too high, but it is not evidence for any issue with a 30/70 allocation.

Nah it is evidence that the vanguard simulator is horribly conservative:) The issue with 30/70 is when you live an extra 5 years much less 10. If your a 60 year old that plans to be dead by 90 and end up making it to 105, being conservative costs you. And of course we are ignoring the fact that the 70/30 case ends up with higher terminal numbers pretty much all the time. How many people do you know that would like to live more money to charity and kids?

The real problem with the bond tent is that it doesn't help. Getting 1% from your bonds instead of say 0% from you stocks over your first 15 years isn't enough to make a difference in your portfolio survivability. Occasionally it works really well (say 1929 when deflation helped bonds or 2000-9 when rate cuts inflated bonds) but it also doesn't make a difference (1966-1981 when bonds and stocks basically had the same 0% returns) or hurts (any of the countries where bonds underperfomed stocks. And yes bonds loses real purchasing power is a relatively common occurrence).

User avatar
nisiprius
Advisory Board
Posts: 38253
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: Should we discourage the bond tent?

Post by nisiprius » Wed Oct 31, 2018 5:39 am

randomguy wrote:
Wed Oct 31, 2018 12:48 am
...Nah it is evidence that the vanguard simulator is horribly conservative:) ...
I originally discovered this back in 2006-2007 using the Fidelity Retirement Income Planner. I came up a little short using my conservative allocation. The tool itself specifically recommended that I increase my stock allocation to a specific number. I reran using the number suggested by the tool itself--and the tool then came up with almost virtually unchanged numbers for "how long your money will last" and "what is the probability of running out."

I like to cite their tool because it's easy to find, simple to use, allows varying parameters, and many Bogleheads already have the necessary login. But it's not just Vanguard. And I don't see how you can call the tool either conservative or aggressive. It doesn't make any suggestions, it doesn't tell you what to choose for "years in retirement," or withdrawal rate, or acceptable failure rate.
The issue with 30/70 is when you live an extra 5 years much less 10.
The results are not wildly different if you extend "years in retirement." For example:

40 years in retirement at 3.5% withdrawal rate: 30/70 yields an 89% success rate, 70/30 yields 90%.
50 years in retirement and a 3% withdrawal rate, 30/70 yields 91% success and 70/30 yields 93%.

And I personally regard these as being pretty aggressive, risk-tolerant choices with regard to failure rate. It's pretty hard to come up with an objective choice of "acceptable failure rate." Personally--given that I am sure failure rates will be much higher in the real world than when planned, and also given that we all acknowledge that we will make course corrections if we see an obvious shortfall--I would not like to plan for a failure rate higher than that of rolling snake-eyes, i.e. 3%. If we cut the 50 year, 70/30 withdrawal rate down until we get 97% success, that occurs at a 2.4% withdrawal rate:

50 years in retirement, 2.4% withdrawal rate: 30/70 yields 98% success, 70/30 yields 97%.

In short, high stock allocations are like leaving your kids lottery tickets with an excellent chance of a jackpot. You might do it for the kids, but, according to simulators, it isn't going to help you have more financial security in retirement, nor will it help you safely use a much higher withdrawal rate. (And the devil of it is that all these simulations turn out to be very sensitive to withdrawal rate--just a few tenths of a percent separates high failure and low failure, not a cliff edge but definitely a soft shoulder--but in real life you don't know where that edge really is.)

What do you object to in Vanguard's simulation tool? What makes it "horribly conservative?"
To estimate these probabilities, we use a technique called Monte Carlo simulation, which attempts to predict how the future may unfold by looking at a wide variety of potential market scenarios that take fluctuating market returns into account. Instead of basing calculations on just one average rate of return, we generate 100,000 simulations of hypothetical market scenarios, and calculate the impact on your savings during your retirement. Each simulation includes up and down markets of various lengths, intensities, and combinations.

For stock market returns we use the Standard & Poor’s 500 Index from 1926 to 1970, the Dow Jones Wilshire 5000 Index from 1971 through April 2005, and the MSCI US Broad Market Index thereafter. For bond market returns, we use the Standard & Poor’s High Grade Corporate Index from 1926 to 1968, the Citigroup High Grade Index from 1969 to 1972, the Barclays US Long Credit AA Index from 1973 to 1975, and the Barclays Capital US Aggregate Bond Index thereafter. For the returns on short-term reserves (i.e., ‘cash’), we use the Citigroup 3-Month Treasury Bill Index. For inflation, we use the changes in the annual Consumer Price Index from 1926 through last year.

For each year of each simulation, we randomly select one year of stock, bond, and cash returns from our database. Using those values, we calculate what would happen to your savings - subtracting your spending, adjusting for inflation, and adding your investment return. We repeat this process, one year at a time, until the end of your retirement or until your savings are exhausted. The next simulation starts the whole process from the beginning.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.

TN_Boy
Posts: 834
Joined: Sat Jan 17, 2009 12:51 pm

Re: Should we discourage the bond tent?

Post by TN_Boy » Wed Oct 31, 2018 8:31 am

nisiprius wrote:
Wed Oct 31, 2018 5:39 am

What do you object to in Vanguard's simulation tool? What makes it "horribly conservative?"
To estimate these probabilities, we use a technique called Monte Carlo simulation, which attempts to predict how the future may unfold by looking at a wide variety of potential market scenarios that take fluctuating market returns into account. Instead of basing calculations on just one average rate of return, we generate 100,000 simulations of hypothetical market scenarios, and calculate the impact on your savings during your retirement. Each simulation includes up and down markets of various lengths, intensities, and combinations.

For stock market returns we use the Standard & Poor’s 500 Index from 1926 to 1970, the Dow Jones Wilshire 5000 Index from 1971 through April 2005, and the MSCI US Broad Market Index thereafter. For bond market returns, we use the Standard & Poor’s High Grade Corporate Index from 1926 to 1968, the Citigroup High Grade Index from 1969 to 1972, the Barclays US Long Credit AA Index from 1973 to 1975, and the Barclays Capital US Aggregate Bond Index thereafter. For the returns on short-term reserves (i.e., ‘cash’), we use the Citigroup 3-Month Treasury Bill Index. For inflation, we use the changes in the annual Consumer Price Index from 1926 through last year.

For each year of each simulation, we randomly select one year of stock, bond, and cash returns from our database. Using those values, we calculate what would happen to your savings - subtracting your spending, adjusting for inflation, and adding your investment return. We repeat this process, one year at a time, until the end of your retirement or until your savings are exhausted. The next simulation starts the whole process from the beginning.
Doesn't the quote from Vanguard potentially answer your question? Here is the key phrase and the reason some people don't like Monte Carlo simulations:

"For each year of each simulation, we randomly select one year of stock, bond, and cash returns from our database"

But nobody believes market returns in year X + 1 are truly independent of returns in year X. The quote above implies the simulation might model sequences like "stocks drop 50%, followed by stocks drop 50%, followed by stock drop 50% ...." etc. And vice versa, of course -- the simulation would have some equally unlikely good returns.

If you believe there is some mean reversion in stock market returns, it seems at least possible that a monte carlo simulation could have more "worst cases" than will happen in real life. Or the worst cases are .... too extreme. I frankly trust historical data more than monte carlo simulations (despite the obvious fact that history might NOT repeat itself) because the historical record has some awful periods, but is also displaying the behavior of real, non-independent, market returns.

What I have not seen is a careful comparison of withdrawal methods using monte carlo and historical returns and a discussion of different results using different simulations types (i.e. I modeled 4% or VPW or whatever using historical data and then I used monte carlo and here are the differences in outcomes based solely in differing simulation methods).

Topic Author
scout1
Posts: 37
Joined: Thu Oct 18, 2018 3:26 pm

Re: Should we discourage the bond tent?

Post by scout1 » Wed Oct 31, 2018 8:45 am

In order to discuss the merits of the bond tent we should establish what the bond tent is. First of the all, the Bond Tent is not the same being overweight bonds before social security hits. That is just proper asset allocation and normal asset/liability matching. Kitces does not mention social security in his article because that is not what the bond tent is about. The bond tent is a strategy to mitigate sequence of return risk that arises because of the “size effect”. The size effect is a behavioral bias because a 50% loss on a large portfolio feels more meaningful than a 50% loss on a small portfolio, even though that is not inherently the case. Because the perceived problem that Kitces is addressing doesn’t actually exist (the size effect), the bond tent strategy just moves people away from a proper asset allocation.

If you have a lower risk tolerance and want to reduce sequence risk you should stay with the monotonically decreasing equity exposure but increase your bond allocation across every single year. If a standard portfolio starts retirement at 50/50 and goes to 10/90, you start yours 45/55 and goe to 5/95. It doesn’t make sense to think you will have a rising equity glide path at any point because you’re always exposed to longevity risk. Just because you’re 85 with a $10m portfolio, you shouldn't start increasing your stock allocation because you could live to be 100+ and with a bear market not have time to make it up.

A question that doesn’t seem to be addressed: What are the characteristics of the person/portfolio who would benefit from such a strategy? If you have a large portfolio with a lower risk tolerance, (which seems like the people who it most appeals to) as I stated above, you should overweight bonds but continue with the normal monotonically decreasing equity allocation.

User avatar
HomerJ
Posts: 12842
Joined: Fri Jun 06, 2008 12:50 pm

Re: Should we discourage the bond tent?

Post by HomerJ » Wed Oct 31, 2018 9:02 am

TN_Boy wrote:
Wed Oct 31, 2018 8:31 am
Doesn't the quote from Vanguard potentially answer your question? Here is the key phrase and the reason some people don't like Monte Carlo simulations:

"For each year of each simulation, we randomly select one year of stock, bond, and cash returns from our database"

But nobody believes market returns in year X + 1 are truly independent of returns in year X.
I've never understood why people like Monte Carlo.

I understand that our historical data set is very limited, and that's a real problem. But random numbers are worse, especially when people are trying to get to 98% "success".

Stock market yearly returns do not appear to be independent returns. People who believe in valuations and "reversion to the mean" and bear\bull cycles should not be using Monte Carlo simulations.
The J stands for Jay

jwhitaker
Posts: 56
Joined: Thu Jun 16, 2016 12:57 pm

Re: Should we discourage the bond tent?

Post by jwhitaker » Wed Oct 31, 2018 9:13 am

I have not done a ton of research on the bond tent, but the whole idea struck me as not making sense. Let's say we have 3 people, all age 65, all have 1M dollars, same spending, etc. etc. Financially indistinguishable. But let's say person 1 has been retired for 10 years, person 2 retired for 5 years and person 3 retires today. A bond tent strategy would imply that these 3 people need to have different bond allocations, but since they are financially indistinguishable that doesn't make sense. Your bond allocation should only depend on current and future conditions. If the strategy doesn't imply that these people have should have different bond allocations then it isn't really a tent right?

ionball
Posts: 153
Joined: Wed Jan 31, 2018 12:17 pm

Re: Should we discourage the bond tent?

Post by ionball » Wed Oct 31, 2018 9:17 am

scout1 wrote:
Wed Oct 31, 2018 8:45 am
What are the characteristics of the person/portfolio who would benefit from such a strategy?
Here's a candidate. Person who thinks sequence of return risk is a valid concern, and needs to match liabilities that are largest at the beginning of retirement period with liability matching amounts projected to decline later in lifecycle. Would the bond tent approach benefit this person or would it prove to be harmful?

Topic Author
scout1
Posts: 37
Joined: Thu Oct 18, 2018 3:26 pm

Re: Should we discourage the bond tent?

Post by scout1 » Wed Oct 31, 2018 9:28 am

ionball wrote:
Wed Oct 31, 2018 9:17 am
scout1 wrote:
Wed Oct 31, 2018 8:45 am
What are the characteristics of the person/portfolio who would benefit from such a strategy?
Here's a candidate. Person who thinks sequence of return risk is a valid concern, and needs to match liabilities that are largest at the beginning of retirement period with liability matching amounts projected to decline later in lifecycle. Would the bond tent approach benefit this person or would it prove to be harmful?
That's still a 100% asset/liability matched portfolio which I agree is a great idea but not the bond tent strategy. The bond tent strategy takes a 100% asset/liability matched portfolio and overweights bonds at the beginning because the "size effect" is larger at the beginning. You can justify any arbitrary glide path if you know exactly when you will have large liabilities, but it's still asset liability matched so it's fine. The bond tent is inherently asset/liability mismatched because the "size effect" exists at the beginning of the portfolio's life even though the size effect is a behavioral bias.

aristotelian
Posts: 5675
Joined: Wed Jan 11, 2017 8:05 pm

Re: Should we discourage the bond tent?

Post by aristotelian » Wed Oct 31, 2018 9:50 am

OP, I wonder if it would help you to think about bond allocation in terms of years expenses rather than %. Once you are retired, the younger you are, the more years expenses you want to have to wait out a market crash and protect you from a poor sequence of returns. If I retire at 62 with 25X expenses, a 50/50 portfolio gives me 12 years expenses, more than enough to last until age 70 and claim SS.

Let's say your portfolio has grown by 20% real to 30X expenses. If you still wanted to have 12 years expenses in bonds, you could now put 18 years in stocks for a 60% allocation. By %, you are more aggressive with a rising equity glide path, but in nominal terms you are still protected (plus you now have SS).

At that point, I would not advise anyone against investing any gains in stocks. As long as you keep in bonds what you can't afford to lose, what you do with the rest is personal preference.

User avatar
David Jay
Posts: 6549
Joined: Mon Mar 30, 2015 5:54 am
Location: Michigan

Re: Should we discourage the bond tent?

Post by David Jay » Wed Oct 31, 2018 9:57 am

I am not doing a 30/70 as in the Kitces example (note that Kitces only introduces it as a concept, he specifically says more research is necessary). My case is that I was 58 years old with a 100% stock portfolio and no pension. I needed a way to fund 8 years of living expenses prior to start of SS without SOR risk. After start of SS, all normal living expenses are covered.

I was looking at the sequence of return risk (more so than portfolio size risk, Kitces other argument). I chose to move all living expenses before SS into bonds. This created about a 45/55 portfolio. I will spend bonds for living expenses and gradually move my AA up to 60/40 at start of SS.

I found the Kitces article as I was implementing my plan, it dovetails nicely with my thinking (great minds... :wink:).
Prediction is very difficult, especially about the future - Niels Bohr | To get the "risk premium", you really do have to take the risk - nisiprius

b0B
Posts: 513
Joined: Fri Oct 19, 2018 11:39 am

Re: Should we discourage the bond tent?

Post by b0B » Wed Oct 31, 2018 10:26 am

My earlier comment might seem jokey and non-serious, but it points out exactly what I could (for)see happening in this thread.

If you construe "Bond Tent" narrowly to mean “Post-Retirement Pre-Determined Rising Equity Glidepath”, then it is indefensible and easily criticized, and the OP is right.

If you construe "Bond Tent" broadly to mean any of a whole wide range of strategies that in some circumstances might kinda sorta vaguely resemble some kind of "Bond Tent"-y thingy, at least some of the time, then you can't out of hand dismiss all such strategies. (But then there's no real reason for the term "Bond Tent" to exist.)

People are taking "sides" based on whether they construe "Bond Tent" narrowly or broadly, rather than based on opinions about retirement investment strategies.
b0B wrote:
Tue Oct 30, 2018 11:57 pm
The “Rising Equity Glidepath” is a very silly idea. The main silliness is robotically following a predetermined glide path regardless of whatever circumstances arise. (And yes, most people wouldn't do that, but computers testing models do exactly that.)

But "Bond Tent" seems a much more nebulous term, so it's harder to pin down for faults.

It's the kind of fuzzy term where two reasonable people could have a conversation:

"Oh I see what you're doing. You're doing a Bond Tent"
"No I'm not. I'm not doing a Bond Tent at all."
"Oh yes you are.
"Oh no I'm not."
"Oh yes you are.
"Oh no I'm not."
etc...

Mavmoses07
Posts: 18
Joined: Sun May 07, 2017 8:37 am
Location: Denver, CO

Re: Should we discourage the bond tent?

Post by Mavmoses07 » Wed Oct 31, 2018 10:28 am

I haven't read through all the comments, so I might have missed something, but it doesn't seem like anyone has provided any empirical evidence to support a more conservative glide path than the bond tent (besides the retirement calculators that use questionable input and Monte Carlo simulations). Has anyone read the research done at earlyretirementnow on this topic?

https://earlyretirementnow.com/2017/09/ ... lidepaths/
https://earlyretirementnow.com/2016/12/ ... t-1-intro/

The first link provides more detailed research into SWRs with various stock/bond allocation (fixed and variable) but with a 60-year time horizon. The second link may be more applicable as shorter time horizons are considered. All the research points to a more aggressive increasing stock allocation in retirement, and it doesn't even include SS and pensions which could take the place of bonds.

What are the counter arguments to this? Is there something fundamental missing from this research?

b0B
Posts: 513
Joined: Fri Oct 19, 2018 11:39 am

Re: Should we discourage the bond tent?

Post by b0B » Wed Oct 31, 2018 10:37 am

^ Any analysis considering only pre-determined glidepaths is too narrow. Glidepaths are for funds not individuals.

Mavmoses07
Posts: 18
Joined: Sun May 07, 2017 8:37 am
Location: Denver, CO

Re: Should we discourage the bond tent?

Post by Mavmoses07 » Wed Oct 31, 2018 10:49 am

b0B wrote:
Wed Oct 31, 2018 10:37 am
^ Any analysis considering only pre-determined glidepaths is too narrow. Glidepaths are for funds not individuals.
Care to elaborate? Isn't that what a bond tent is, a glidepath? Are you suggesting one sticks with one constant stock/bond allocation post retirement? Unless I'm missing something, none of that invalids the research.

User avatar
willthrill81
Posts: 10122
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: Should we discourage the bond tent?

Post by willthrill81 » Wed Oct 31, 2018 10:52 am

HomerJ wrote:
Wed Oct 31, 2018 9:02 am
TN_Boy wrote:
Wed Oct 31, 2018 8:31 am
Doesn't the quote from Vanguard potentially answer your question? Here is the key phrase and the reason some people don't like Monte Carlo simulations:

"For each year of each simulation, we randomly select one year of stock, bond, and cash returns from our database"

But nobody believes market returns in year X + 1 are truly independent of returns in year X.
I've never understood why people like Monte Carlo.

I understand that our historical data set is very limited, and that's a real problem. But random numbers are worse, especially when people are trying to get to 98% "success".

Stock market yearly returns do not appear to be independent returns. People who believe in valuations and "reversion to the mean" and bear\bull cycles should not be using Monte Carlo simulations.
:thumbsup

It has already been shown that Monte Carlo simulations that do not specifically incorporate mean reversion produce significantly more extreme sequences on both the negative and positive side than the historical record, more than what would be expected statistically. According to Occam's razor, the most logical conclusion is that such MC simulations are flawed.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

Topic Author
scout1
Posts: 37
Joined: Thu Oct 18, 2018 3:26 pm

Re: Should we discourage the bond tent?

Post by scout1 » Wed Oct 31, 2018 10:55 am

Mavmoses07 wrote:
Wed Oct 31, 2018 10:28 am
I haven't read through all the comments, so I might have missed something, but it doesn't seem like anyone has provided any empirical evidence to support a more conservative glide path than the bond tent (besides the retirement calculators that use questionable input and Monte Carlo simulations). Has anyone read the research done at earlyretirementnow on this topic?

https://earlyretirementnow.com/2017/09/ ... lidepaths/
https://earlyretirementnow.com/2016/12/ ... t-1-intro/

The first link provides more detailed research into SWRs with various stock/bond allocation (fixed and variable) but with a 60-year time horizon. The second link may be more applicable as shorter time horizons are considered. All the research points to a more aggressive increasing stock allocation in retirement, and it doesn't even include SS and pensions which could take the place of bonds.

What are the counter arguments to this? Is there something fundamental missing from this research?
Huge problems here, they are completely, 100% wrong.

They start off by assuming Kitces is wrong and saying Vanguard got it wrong (Kitces got it wrong). As a result of this assumption they ONLY test rising equity glide paths which doesn't even make sense. Why would a 90 year old at 100% stocks be an optimal allocation? They could still live another 15 years and need the money.
Last edited by scout1 on Wed Oct 31, 2018 11:07 am, edited 4 times in total.

b0B
Posts: 513
Joined: Fri Oct 19, 2018 11:39 am

Re: Should we discourage the bond tent?

Post by b0B » Wed Oct 31, 2018 11:02 am

Mavmoses07 wrote:
Wed Oct 31, 2018 10:49 am
b0B wrote:
Wed Oct 31, 2018 10:37 am
^ Any analysis considering only pre-determined glidepaths is too narrow. Glidepaths are for funds not individuals.
Care to elaborate? Isn't that what a bond tent is, a glidepath? Are you suggesting one sticks with one constant stock/bond allocation post retirement? Unless I'm missing something, none of that invalids the research.
Assuming that a strategy must take the form of a pre-determined glidepath is a fatal flaw. It's okay to do that analysis, but be aware that it is a huge oversimplification - basically just a toy model.

ionball
Posts: 153
Joined: Wed Jan 31, 2018 12:17 pm

Re: Should we discourage the bond tent?

Post by ionball » Wed Oct 31, 2018 11:23 am

scout1 wrote:
Wed Oct 31, 2018 10:55 am
Why would a 90 year old at 100% stocks be an optimal allocation?
There are a variety of 90 year olds. Some may not need to withdraw from their portfolio. If they hold the portfolio for long term legacy purposes, and the portfolio is in taxable account, and they want to minimize income taxation, then they may consider 100% stocks appropriate. Optimal is unlikely to be achieved without hindsight.

This is only one case to consider. Many other cases would be better off with a different allocation.

Perhaps the work by Kitces was flawed, but I found it useful and thought provoking. I would not discourage a bond tent in all cases. Nor would I discourage a bond tent for an individual case unless I was convinced that it could cause damage. Just a point of view in the fog.

Mavmoses07
Posts: 18
Joined: Sun May 07, 2017 8:37 am
Location: Denver, CO

Re: Should we discourage the bond tent?

Post by Mavmoses07 » Wed Oct 31, 2018 11:24 am

scout1 wrote:
Wed Oct 31, 2018 10:55 am
Huge problems here, they are completely, 100% wrong.

They start off by assuming Kitces is wrong and saying Vanguard got it wrong (Kitces got it wrong). As a result of this assumption they ONLY test rising equity glide paths which doesn't even make sense. Why would a 90 year old at 100% stocks be an optimal allocation? They could still live another 15 years and need the money.
For a long time horizon one needs to count on the greater returns of stocks vs. lower returns of bonds in the long-run. As time horizon increases, the portfolios with greater stock allocations beat those with greater bond allocations (at least based on the past ~150 years of stock/bond returns). Though I agree, I can't imagine a 90-year-old being 100% invested in stocks.

He's done additional research on static stock/bond allocations, see second link. How would you analyze it? How would you adjust your stock/bond allocation, how would you justify it? Just because he chose a stock/bond allocation it invalids all the research?!

Topic Author
scout1
Posts: 37
Joined: Thu Oct 18, 2018 3:26 pm

Re: Should we discourage the bond tent?

Post by scout1 » Wed Oct 31, 2018 11:30 am

Mavmoses07 wrote:
Wed Oct 31, 2018 11:24 am
Huge problems here, they are completely, 100% wrong.

They start off by assuming Kitces is wrong and saying Vanguard got it wrong (Kitces got it wrong). As a result of this assumption they ONLY test rising equity glide paths which doesn't even make sense. Why would a 90 year old at 100% stocks be an optimal allocation? They could still live another 15 years and need the money.
For a long time horizon one needs to count on the greater returns of stocks vs. lower returns of bonds in the long-run. As time horizon increases, the portfolios with greater stock allocations beat those with greater bond allocations (at least based on the past ~150 years of stock/bond returns). Though I agree, I can't imagine a 90-year-old being 100% invested in stocks.

He's done additional research on static stock/bond allocations, see second link. How would you analyze it? How would you adjust your stock/bond allocation, how would you justify it? Just because he chose a stock/bond allocation it invalids all the research?!
They just need to compare what they have tested to declining equity glide path and the declining one will look superior to what they have shown in their tables. That's standard stuff and what every target date fund does. Target date funds don't lower the equity portion over time just so the retireee can jump back into equities when they're 85.

With respect to static S/B allocations, if your time horizon is 60 years of course stocks will look great (and they are great). That's why every 30 year old is 100% stocks, they have a 60 year horizon.

Mavmoses07
Posts: 18
Joined: Sun May 07, 2017 8:37 am
Location: Denver, CO

Re: Should we discourage the bond tent?

Post by Mavmoses07 » Wed Oct 31, 2018 11:48 am

scout1 wrote:
Wed Oct 31, 2018 11:30 am
They just need to compare what they have tested to declining equity glide path and the declining one will look superior to what they have shown in their tables...
Declining equity paths how a lower SWR than other portfolios for a 30-year time horizon, at least according to Cfiresim and other retirement calculators that use historical data. If you know of research to contrary I'd love to see it.
scout1 wrote:
Wed Oct 31, 2018 11:30 am
With respect to static S/B allocations, if your time horizon is 60 years of course stocks will look great. That's why every 30 year old is 100% stocks, they have a 60 year horizon.
His research is for post-retirement. He looked at 30 year time horizons as well and stock portfolios were shown to be superior.

TN_Boy
Posts: 834
Joined: Sat Jan 17, 2009 12:51 pm

Re: Should we discourage the bond tent?

Post by TN_Boy » Wed Oct 31, 2018 11:55 am

scout1 wrote:
Wed Oct 31, 2018 11:30 am
Mavmoses07 wrote:
Wed Oct 31, 2018 11:24 am
Huge problems here, they are completely, 100% wrong.

They start off by assuming Kitces is wrong and saying Vanguard got it wrong (Kitces got it wrong). As a result of this assumption they ONLY test rising equity glide paths which doesn't even make sense. Why would a 90 year old at 100% stocks be an optimal allocation? They could still live another 15 years and need the money.
For a long time horizon one needs to count on the greater returns of stocks vs. lower returns of bonds in the long-run. As time horizon increases, the portfolios with greater stock allocations beat those with greater bond allocations (at least based on the past ~150 years of stock/bond returns). Though I agree, I can't imagine a 90-year-old being 100% invested in stocks.

He's done additional research on static stock/bond allocations, see second link. How would you analyze it? How would you adjust your stock/bond allocation, how would you justify it? Just because he chose a stock/bond allocation it invalids all the research?!
They just need to compare what they have tested to declining equity glide path and the declining one will look superior to what they have shown in their tables. That's standard stuff and what every target date fund does. Target date funds don't lower the equity portion over time just so the retireee can jump back into equities when they're 85.

With respect to static S/B allocations, if your time horizon is 60 years of course stocks will look great (and they are great). That's why every 30 year old is 100% stocks, they have a 60 year horizon.
scout1,

Your posts seem to be "I think the bond tent is wrong." Could you cite a reference with some backtesting rather than just assert "XXX is wrong?"

What data/studies/papers have you seen convincing you this is true:

"They just need to compare what they have tested to declining equity glide path and the declining one will look superior to what they have shown in their tables."

I look at the ERN post on rising glidepaths and find it makes a decent case for rising glidepaths, especially (a case that interests me) if you retire when equity valuations are high. Also, a couple of people have pointed out how it may fit nicely with a scenario where you live off portfolio and take SS later -- your bond percentage is higher during those "bridge" years.

I don't think there is great evidence for any strategy being superior (as we do not know the future) but I am looking for what particular data/study has so convinced you the rising equity glidepath is obviously wrong.

I'm also going to make this point again, which I think you are overlooking:

An 85 year old may or may not have a 10 year horizon. I know a number of people who were like, dead, at 85. They didn't need any money at all. And even if they were alive, you need a lot less money to handle 10 years than 30.

It's important; to be poor at 85 when your retirement finances were reasonable, not one, but two things had to happen.

1) you had awful market returns (or exceptional expenses) and
2) you have to be alive.

User avatar
David Jay
Posts: 6549
Joined: Mon Mar 30, 2015 5:54 am
Location: Michigan

Re: Should we discourage the bond tent?

Post by David Jay » Wed Oct 31, 2018 11:59 am

scout1 wrote:
Wed Oct 31, 2018 11:30 am
They just need to compare what they have tested to declining equity glide path and the declining one will look superior to what they have shown in their tables. That's standard stuff and what every target date fund does.
Help me understand. You seem to be saying that a portfolio with a declining equity position will outperform a portfolio with an increasing equity position over the long term (say 30 years)?
Prediction is very difficult, especially about the future - Niels Bohr | To get the "risk premium", you really do have to take the risk - nisiprius

Topic Author
scout1
Posts: 37
Joined: Thu Oct 18, 2018 3:26 pm

Re: Should we discourage the bond tent?

Post by scout1 » Wed Oct 31, 2018 12:01 pm

Mavmoses07 wrote:
Wed Oct 31, 2018 11:48 am
scout1 wrote:
Wed Oct 31, 2018 11:30 am
They just need to compare what they have tested to declining equity glide path and the declining one will look superior to what they have shown in their tables...
Declining equity paths how a lower SWR than other portfolios for a 30-year time horizon, at least according to Cfiresim and other retirement calculators that use historical data. If you know of research to contrary I'd love to see it.
no they don't... I just went to cFIREsim:

2018 - 2047 simulation
2018-2033 is 75/25 stock/bond
2033-2047 is 25/75.
Success rate is 95%.

If you reverse that to 25/75 stock bond in 2018 and then 75/25 from 2033-2047 the success rate is 78%.
Mavmoses07 wrote:
Wed Oct 31, 2018 11:48 am
scout1 wrote:
Wed Oct 31, 2018 11:30 am
With respect to static S/B allocations, if your time horizon is 60 years of course stocks will look great. That's why every 30 year old is 100% stocks, they have a 60 year horizon.

His research is for post-retirement. He looked at 30 year time horizons as well and stock portfolios were shown to be superior.
I agree stocks are great, which is why you shouldn't overweight bonds when you're just starting retirement like the bond tent suggests.
David Jay wrote:
Wed Oct 31, 2018 11:59 am
scout1 wrote:
Wed Oct 31, 2018 11:30 am
They just need to compare what they have tested to declining equity glide path and the declining one will look superior to what they have shown in their tables. That's standard stuff and what every target date fund does.
Help me understand. You seem to be saying that a portfolio with a declining equity position will outperform a portfolio with an increasing equity position over the long term (say 30 years)?
Yes. A declining path is optimal and that's why target date funds use declining paths and not rising paths and for funds 30 years out.

megabad
Posts: 1740
Joined: Fri Jun 01, 2018 4:00 pm

Re: Should we discourage the bond tent?

Post by megabad » Wed Oct 31, 2018 12:03 pm

scout1 wrote:
Mon Oct 29, 2018 3:09 pm
To me it looks like a bad strategy and I believe BH'ers should push back on it.
I have a hard time predicting what will be a good or bad strategy in the future.

The bond tent violates the most fundamental rule of asset allocation: money should be allocated to match time horizon.
I do not interpret the "bond tent" to be an investment strategy, but rather a form of thought theory. Kitces appears to suggest that there is an exact fixed portfolio size required at retirement and that a subject can choose his/her exact retirement date and push this date off if a portfolio is not "large enough". These are very exact ways of thinking about the future that I don't subscribe to, but I wouldn't criticize people for taking this view (nor do I find it "bad"). I have a much less black & white and much grayer view of the future. I will retire when I feel like it, or when I am forced to (medically or by employer). Neither of these things involve how I invest or exact numbers/dates. If my portfolio takes a hit, I will live on less.

The appeal of the strategy seems to be emotional since people get to overweight bonds when they have a large portfolio.
Well since I believe this is a just a way of looking at retirement planning, I would agree that it is emotional. It does not necessarily indicate that anyone would be "overweight" in bonds (What is overweight?). This is no different in my view than the "liability matching" and the "bucket planning" approaches. Neither necessarily controls your financial strategy, but are rather just ways of looking at the situation. A 50/50 portfolio is still a 50/50 portfolio.

User avatar
David Jay
Posts: 6549
Joined: Mon Mar 30, 2015 5:54 am
Location: Michigan

Re: Should we discourage the bond tent?

Post by David Jay » Wed Oct 31, 2018 12:24 pm

scout1 wrote:
Wed Oct 31, 2018 12:01 pm
David Jay wrote:
Wed Oct 31, 2018 11:59 am
scout1 wrote:
Wed Oct 31, 2018 11:30 am
They just need to compare what they have tested to declining equity glide path and the declining one will look superior to what they have shown in their tables. That's standard stuff and what every target date fund does.
Help me understand. You seem to be saying that a portfolio with a declining equity position will outperform a portfolio with an increasing equity position over the long term (say 30 years)?
Yes. A declining path is optimal and that's why target date funds use declining paths and not rising paths and for funds 30 years out.
Could you point out a target date fund that has a decreasing equity glideslope through the retirement years? The ones I have seen (i.e. VG TargetRetirement, Fid. Freedom Index) hold a fixed allocation in the retirement years.
Prediction is very difficult, especially about the future - Niels Bohr | To get the "risk premium", you really do have to take the risk - nisiprius

User avatar
HomerJ
Posts: 12842
Joined: Fri Jun 06, 2008 12:50 pm

Re: Should we discourage the bond tent?

Post by HomerJ » Wed Oct 31, 2018 12:31 pm

scout1 wrote:
Wed Oct 31, 2018 10:55 am
they are completely, 100% wrong.
Again with the 100% statements.

A kindly note to a brand-new poster.... First, welcome!

Second, I would suggest that you can disagree without being so vehement. Especially when you're brand new and just moved into the neighborhood.
The J stands for Jay

Topic Author
scout1
Posts: 37
Joined: Thu Oct 18, 2018 3:26 pm

Re: Should we discourage the bond tent?

Post by scout1 » Wed Oct 31, 2018 12:37 pm

TN_Boy wrote:
Wed Oct 31, 2018 11:55 am

scout1,

Your posts seem to be "I think the bond tent is wrong." Could you cite a reference with some backtesting rather than just assert "XXX is wrong?"

What data/studies/papers have you seen convincing you this is true:

"They just need to compare what they have tested to declining equity glide path and the declining one will look superior to what they have shown in their tables."

I look at the ERN post on rising glidepaths and find it makes a decent case for rising glidepaths, especially (a case that interests me) if you retire when equity valuations are high. Also, a couple of people have pointed out how it may fit nicely with a scenario where you live off portfolio and take SS later -- your bond percentage is higher during those "bridge" years.
Market timing is great if you can do it. I agree SS needs to be factored in an so you do end up overweighting bonds prior to SS but that's not what a bond tent is for. The bond tent is for mitigating a false fear Kitces refers to as "the portfolio size effect"
TN_Boy wrote:
Wed Oct 31, 2018 11:55 am
I don't think there is great evidence for any strategy being superior (as we do not know the future) but I am looking for what particular data/study has so convinced you the rising equity glidepath is obviously wrong.
put a rising glide path into cFIREsim. Your success rate goes down.
TN_Boy wrote:
Wed Oct 31, 2018 11:55 am
I'm also going to make this point again, which I think you are overlooking:

An 85 year old may or may not have a 10 year horizon. I know a number of people who were like, dead, at 85. They didn't need any money at all. And even if they were alive, you need a lot less money to handle 10 years than 30.

It's important; to be poor at 85 when your retirement finances were reasonable, not one, but two things had to happen.

1) you had awful market returns (or exceptional expenses) and
2) you have to be alive.
you don't realistically get to pick when you die. Heck if you're on the verge of death since you're health is so bad and you're getting lots of healthcare treatment you probably need even more bonds, not fewer.

Also, yes you need fewer years but more in bonds relative to stocks.
David Jay wrote:
Wed Oct 31, 2018 12:24 pm

Could you point out a target date fund that has a decreasing equity glideslope through the retirement years? The ones I have seen (i.e. VG TargetRetirement, Fid. Freedom Index) hold a fixed allocation in the retirement years.
Many of them do.

https://www.hightoweradvisors.com/team/ ... ifferences
megabad wrote:
Wed Oct 31, 2018 12:03 pm

(What is overweight?).
Overweight is simply relative to an asset/liability matched portfolio. Bond tent practitioners emphasize bonds early in retirement because they are concerned about the "size effect" but my argument is the size effect isn't real so the portfolio is just worse.

TN_Boy
Posts: 834
Joined: Sat Jan 17, 2009 12:51 pm

Re: Should we discourage the bond tent?

Post by TN_Boy » Wed Oct 31, 2018 1:40 pm

scout1 wrote:
Wed Oct 31, 2018 12:37 pm
TN_Boy wrote:
Wed Oct 31, 2018 11:55 am

scout1,

Your posts seem to be "I think the bond tent is wrong." Could you cite a reference with some backtesting rather than just assert "XXX is wrong?"

What data/studies/papers have you seen convincing you this is true:

"They just need to compare what they have tested to declining equity glide path and the declining one will look superior to what they have shown in their tables."

I look at the ERN post on rising glidepaths and find it makes a decent case for rising glidepaths, especially (a case that interests me) if you retire when equity valuations are high. Also, a couple of people have pointed out how it may fit nicely with a scenario where you live off portfolio and take SS later -- your bond percentage is higher during those "bridge" years.
Market timing is great if you can do it. I agree SS needs to be factored in an so you do end up overweighting bonds prior to SS but that's not what a bond tent is for. The bond tent is for mitigating an false fear Kitces refers to as "the portfolio size effect"
TN_Boy wrote:
Wed Oct 31, 2018 11:55 am
I don't think there is great evidence for any strategy being superior (as we do not know the future) but I am looking for what particular data/study has so convinced you the rising equity glidepath is obviously wrong.
put a rising glide path into cFIREsim. Your success rate goes down.
TN_Boy wrote:
Wed Oct 31, 2018 11:55 am
I'm also going to make this point again, which I think you are overlooking:

An 85 year old may or may not have a 10 year horizon. I know a number of people who were like, dead, at 85. They didn't need any money at all. And even if they were alive, you need a lot less money to handle 10 years than 30.

It's important; to be poor at 85 when your retirement finances were reasonable, not one, but two things had to happen.

1) you had awful market returns (or exceptional expenses) and
2) you have to be alive.
you don't realistically get to pick when you die. Heck if you're on the verge of death since you're health is so bad and you're getting lots of healthcare treatment you probably need even more bonds, not fewer.

Also, yes you need fewer years but more in bonds relative to stocks.
1) I think when you used cFireSim, you did two different simulations, one with 25% stocks, then after 15 years another with 75% stocks? This is not how the rising glidepath works; it starts low(er) and rises every year. It's not a step function. Also, a higher starting percentage than 25% is what most people recommend (see the ERN work).

This is what you said:

2018 - 2047 simulation
2018-2033 is 75/25 stock/bond
2033-2047 is 25/75.

But maybe that is not what you meant to say? When I get a chance, I'll also run some cFiresim simulations.

2) I am not a fan of market timing at all, though multiple researchers have found some value in using CAPE. But that's a somewhat valid criticism of my comment.

3) At 85, there is an outstanding chance you will need money for only a small number of years, hence the exact allocation is less important.

User avatar
David Jay
Posts: 6549
Joined: Mon Mar 30, 2015 5:54 am
Location: Michigan

Re: Should we discourage the bond tent?

Post by David Jay » Wed Oct 31, 2018 1:43 pm

scout1 wrote:
Wed Oct 31, 2018 12:37 pm
David Jay wrote:
Wed Oct 31, 2018 12:24 pm

Could you point out a target date fund that has a decreasing equity glideslope through the retirement years? The ones I have seen (i.e. VG TargetRetirement, Fid. Freedom Index) hold a fixed allocation in the retirement years.
Many of them do.

https://www.hightoweradvisors.com/team/ ... ifferences
Interesting reference. It shows that two (of the thirteen listed) do not reach terminal asset allocations, or at least not in the 30 years after retirement. I was not aware of that.
Prediction is very difficult, especially about the future - Niels Bohr | To get the "risk premium", you really do have to take the risk - nisiprius

Mavmoses07
Posts: 18
Joined: Sun May 07, 2017 8:37 am
Location: Denver, CO

Re: Should we discourage the bond tent?

Post by Mavmoses07 » Wed Oct 31, 2018 1:56 pm

scout1 wrote:
Wed Oct 31, 2018 12:01 pm

no they don't... I just went to cFIREsim:

2018 - 2047 simulation
2018-2033 is 75/25 stock/bond
2033-2047 is 25/75.
Success rate is 95%.

If you reverse that to 25/75 stock bond in 2018 and then 75/25 from 2033-2047 the success rate is 78%.
And a 75% static stock allocation and a 60/40 ==>100/0 glidepath do better than both. I'll also note that failure means having at least $1 after 30 years. If you revise your goal to have 50% or 100% of what you started with, the higher stock allocations perform even better than the bond heavy portfolios.

randomguy
Posts: 7676
Joined: Wed Sep 17, 2014 9:00 am

Re: Should we discourage the bond tent?

Post by randomguy » Wed Oct 31, 2018 1:58 pm

nisiprius wrote:
Wed Oct 31, 2018 5:39 am
randomguy wrote:
Wed Oct 31, 2018 12:48 am
...Nah it is evidence that the vanguard simulator is horribly conservative:) ...
I originally discovered this back in 2006-2007 using the Fidelity Retirement Income Planner. I came up a little short using my conservative allocation. The tool itself specifically recommended that I increase my stock allocation to a specific number. I reran using the number suggested by the tool itself--and the tool then came up with almost virtually unchanged numbers for "how long your money will last" and "what is the probability of running out."

I like to cite their tool because it's easy to find, simple to use, allows varying parameters, and many Bogleheads already have the necessary login. But it's not just Vanguard. And I don't see how you can call the tool either conservative or aggressive. It doesn't make any suggestions, it doesn't tell you what to choose for "years in retirement," or withdrawal rate, or acceptable failure rate.
The issue with 30/70 is when you live an extra 5 years much less 10.
The results are not wildly different if you extend "years in retirement." For example:

40 years in retirement at 3.5% withdrawal rate: 30/70 yields an 89% success rate, 70/30 yields 90%.
50 years in retirement and a 3% withdrawal rate, 30/70 yields 91% success and 70/30 yields 93%.

And I personally regard these as being pretty aggressive, risk-tolerant choices with regard to failure rate. It's pretty hard to come up with an objective choice of "acceptable failure rate." Personally--given that I am sure failure rates will be much higher in the real world than when planned, and also given that we all acknowledge that we will make course corrections if we see an obvious shortfall--I would not like to plan for a failure rate higher than that of rolling snake-eyes, i.e. 3%. If we cut the 50 year, 70/30 withdrawal rate down until we get 97% success, that occurs at a 2.4% withdrawal rate:

50 years in retirement, 2.4% withdrawal rate: 30/70 yields 98% success, 70/30 yields 97%.

Real world numbers.

40 years at 3.75%
70/30 94.5%
30/70 64.8%

40 years @3.5%
70/30 99.1%
30/70 87.1%

50 years at 3%
70/30 100%
30/70 89.6%

I consider those a pretty big gap:). Why the difference between that and what vanguard put out. My guess is handling of inflation and mean reversion.

And yes you can keep going hyper conservative until things work out.

Post Reply