SWR with a real return of 1.6%

Non-investing personal finance issues including insurance, credit, real estate, taxes, employment and legal issues such as trusts and wills
Post Reply
Ron Scott
Posts: 339
Joined: Tue Apr 05, 2016 5:38 am

SWR with a real return of 1.6%

Post by Ron Scott » Fri Oct 27, 2017 7:44 pm

I'm inclined to base my financial decisions for the foreseeable future on Bogle's most recent assumptions about future-decade returns. With a 60-40 allocation and an average 2% inflation rate he sees a real rate of return of 1.6%. That compares to 6.3% in the last 45 years or so and 4.8% for the past century; a big drop...but there you have it.

Some of you will certainly disagree with this assessment, which is all good and you'll get no argument from me, but Jack's success or failure to predict the future is not the purpose of this thread. What I'm interested in learning is how real returns affect SWR calculations.

If you believed a 4% SWR in the past was based on a real return of say 4.8%, what math do we use to determine SWR at 1.6% (and what's the answer)?

AlohaJoe
Posts: 2550
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: SWR with a real return of 1.6%

Post by AlohaJoe » Fri Oct 27, 2017 7:56 pm

A 4% SWR wasn't based on a real return of 4.8%.

It is based on volatility and Bogle has never made a prediction about volatility.

Ron Scott
Posts: 339
Joined: Tue Apr 05, 2016 5:38 am

Re: SWR with a real return of 1.6%

Post by Ron Scott » Sat Oct 28, 2017 8:01 am

AlohaJoe wrote:
Fri Oct 27, 2017 7:56 pm
A 4% SWR wasn't based on a real return of 4.8%.

It is based on volatility and Bogle has never made a prediction about volatility.
Well...

It cannot be ALL about volatility. If I put my money under a mattress and live for 30 years off it, inflating the spend by 2% annually and achieving 0% growth in the nest egg, I can start my spend at ~2.45% of the nest egg and run out of money at death. (Zero volatility) We can argue about whether my inflation goal is adequate but I might be willing to live with the consequences regardless.

Safe and perpetual withdrawal rate calculators (like this one: https://portfoliocharts.com/portfolio/withdrawal-rates/) show that taking some risk in stock/bond allocations increases the %s you can safely withdraw, i.e., the more you earn on the nest egg the more you can spend.

Using the calculator I linked too I'll offer my own imperfect answer to my original question by replacing their 60-40 kickstarter with 100% cash, and assume a 30 year lifespan:

At 60-40 the SWR is ~4.5% and the perpetual is about 3.6%.
At 100% cash (overly conservative surrogate for Boggle's prediction) SWR is about 3.6% and perpetual is about 1.6%

Volatility obviously counts, as the 100% stock portfolio does worse than a stock/bond mix, but earnings count too. Question is how much, if one believes Bogle's big-drop prediction.

User avatar
TheTimeLord
Posts: 4657
Joined: Fri Jul 26, 2013 2:05 pm

Re: SWR with a real return of 1.6%

Post by TheTimeLord » Sat Oct 28, 2017 8:13 am

Ron Scott wrote:
Fri Oct 27, 2017 7:44 pm
If you believed a 4% SWR in the past was based on a real return of say 4.8%, what math do we use to determine SWR at 1.6% (and what's the answer)?
Seems incorrect to me to say the SWR for a 4.8% real return is 4.0%. So I think you need to consider other factors in your evaluation. Even if the real return is sub 4.0%, you can maintain a 4.0% WR for quite awhile by reducing your portfolio balance. It seems to me a lot of people here seem to believe they are suppose to die with as much or more money than they had at retirement adjusted for inflation. Maybe that is just their experience with the recent Bull market, maybe it is because it is a simpler calculation to just have just say X dollars at Y real rate gives me Z dollars to live on but I really don't think that is how it works, especially considering market volatility.
Run, You Clever Boy!

AlohaJoe
Posts: 2550
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: SWR with a real return of 1.6%

Post by AlohaJoe » Sat Oct 28, 2017 8:18 am

With 0 volatility a 4% SWR only requires around 1.3% real returns.

So, yes, it is all about volatility. That's why retirement researchers have made such a big deal about "sequence of returns" over the past 30 years.

The 4% SWR is based on a 30 year period (1966-1996) when real returns were way above 1.3% real. But the volatility, 3 crashes in 8 years, is what caused problems.

dbr
Posts: 23687
Joined: Sun Mar 04, 2007 9:50 am

Re: SWR with a real return of 1.6%

Post by dbr » Sat Oct 28, 2017 8:40 am

AlohaJoe wrote:
Sat Oct 28, 2017 8:18 am
With 0 volatility a 4% SWR only requires around 1.3% real returns.

So, yes, it is all about volatility. That's why retirement researchers have made such a big deal about "sequence of returns" over the past 30 years.

The 4% SWR is based on a 30 year period (1966-1996) when real returns were way above 1.3% real. But the volatility, 3 crashes in 8 years, is what caused problems.
Joe is right. You can set up a ladder of 30 year TIPS at 1.3% real and get 4% real back for 30 years. He is also right about volatility. Also, SWR research has used data for much longer periods than 1966-1996. It is not a general truth that higher return portfolios have higher withdrawal rates. Return and variability of return have offsetting effects on SWR and the optimum SWR does not generally occur at 100% stocks.

Aside from volatility Bogle is not predicting 30 year results. He is guessing what is going to happen for the next decade or so. That isn't enough information to decide what is going to happen to someone who retires today. You can try to run a retirement model conditional on thinking you know something about how a retirement today is going to differ from retirements on average but I doubt anyone is going to be very accurate in doing so. It is hard enough to get meaningful numbers at all from average data.

Ron Scott
Posts: 339
Joined: Tue Apr 05, 2016 5:38 am

Re: SWR with a real return of 1.6%

Post by Ron Scott » Sat Oct 28, 2017 9:41 am

Interesting. Then what's the point of taking risk in retirement with stocks, etc.

User avatar
TheTimeLord
Posts: 4657
Joined: Fri Jul 26, 2013 2:05 pm

Re: SWR with a real return of 1.6%

Post by TheTimeLord » Sat Oct 28, 2017 9:50 am

Ron Scott wrote:
Sat Oct 28, 2017 9:41 am
Interesting. Then what's the point of taking risk in retirement with stocks, etc.
Living more than 30 years in retirement?
Run, You Clever Boy!

dbr
Posts: 23687
Joined: Sun Mar 04, 2007 9:50 am

Re: SWR with a real return of 1.6%

Post by dbr » Sat Oct 28, 2017 10:38 am

Ron Scott wrote:
Sat Oct 28, 2017 9:41 am
Interesting. Then what's the point of taking risk in retirement with stocks, etc.
As Timelord said . . . Hopefully you understand that at the end of living off one bond a year maturing and being spent, you now have zero assets and zero income. This is an explicit and definite result intended from the beginning. When people do SWR studies and put on 30 year time spans they are just creating an informative model for how things work and not expecting people to use this as an actual plan. In reality in those plans most of the outcomes don't fail and the retiree still has lots of money. You can run the plan to see what happens at longer time periods. Of course you can buy a TIPS ladder that runs longer (in theory, but in practice 30 years is the longest issue available). A better idea is an inflation indexed annuity, if you want a definite payout in real dollars and lifelong. At low interest rates of today if you can find such an annuity I guess the payout is probably around 4%.

randomguy
Posts: 4871
Joined: Wed Sep 17, 2014 9:00 am

Re: SWR with a real return of 1.6%

Post by randomguy » Sat Oct 28, 2017 10:48 am

AlohaJoe wrote:
Sat Oct 28, 2017 8:18 am
With 0 volatility a 4% SWR only requires around 1.3% real returns.

So, yes, it is all about volatility. That's why retirement researchers have made such a big deal about "sequence of returns" over the past 30 years.

The 4% SWR is based on a 30 year period (1966-1996) when real returns were way above 1.3% real. But the volatility, 3 crashes in 8 years, is what caused problems.
Volatility only caused part of the problems. Rapid devaluation of the currency caused the rest. It was sort of a unque 30 year period. It was just about the worst 15 year period every (-1% real), followed by one of the best (12.49%) for stocks and bonds were pretty much the same. The great depression had worse stock market performances but the fact that bonds didn't crater prevented it from being the rate limiting time period.

As for predicting the future, it all comes down to how we end up with 1.6% for th enext 10 years AND what happens afterwards. 9 years of steady growth and a big correction isn't anywhere near as devasting as a big correction, and slow growth..

User avatar
#Cruncher
Posts: 2418
Joined: Fri May 14, 2010 2:33 am
Location: New York City
Contact:

Re: SWR with a real return of 1.6%

Post by #Cruncher » Sat Oct 28, 2017 10:56 am

TIPS yields aren't quite that high now, dbr. Better to have said:
dbr wrote:
Sat Oct 28, 2017 8:40 am
You can [If you could] set up a ladder of 30 year TIPS at 1.3% real and [you'd] get 4% real back for 30 years.
But your 4% is close anyway. According to my latest TIPS Ladder Builder spreadsheet, with 07/24/2017 prices, $804,000 [*] buys enough TIPS to provide $30,000 in constant dollars for 30 years, a "withdrawal rate" of 3.7%.
Ron Scott wrote:
Sat Oct 28, 2017 9:41 am
Interesting. Then what's the point of taking risk in retirement with stocks, etc.
Assuming you're referring to dbr's TIPS ladder, the reason one might take risk is the potential to have money left over after 30 years. There is no chance of this with the TIPS ladder.

For what it's worth, here is my estimate that if a 4.8% return supports a 4% withdrawal rate, then a 1.6% return would support a 3% withdrawal. This is based on the severe case where the portfolio falls 10% every year for the first five years and then grows at a constant rate for the next 25 years such that the overall 30 year growth rate is 4.8% or 1.6%. Given this assumption, both the 4.8% growth/4% withdrawal and 1.6% growth/3% withdrawal scenarios last 30 years before failing. However, a 4% withdrawal with only 1.6% overall growth would fail after 19 years. (For more on this simplistic methodology, see my post, Re: Avoiding the Spiral.)

Code: Select all

 1 Initial balance 1,000,000 1,000,000 1,000,000
 2 Withdrawal 4.000% 3.000% 4.000% <--
 5 Portfolio average growth 4.800% 1.600% 1.600% <--
 6 Over this many years 30 30 30
 7 But for first [Y] years 5 5 5
 8 It falls [F] (10.000%) (10.000%) (10.000%)
 9 Thereafter it rises [R] 8.040% 4.094% 4.094%
 Year ------------ Balance ------------

Code: Select all

 11 0 1,000,000 1,000,000 1,000,000
 12 1 860,000 870,000 860,000
 13 2 734,000 753,000 734,000
 14 3 620,600 647,700 620,600
 15 4 518,540 552,930 518,540
 16 5 426,686 467,637 426,686
 17 6 420,992 456,780 404,153
 18 7 414,840 445,479 380,697
 19 8 408,194 433,715 356,281
 20 9 401,013 421,469 330,866
 21 10 393,255 408,722 304,410
 22 11 384,873 395,454 276,871
 23 12 375,818 381,642 248,205
 24 13 366,034 367,265 218,366
 25 14 355,463 352,299 187,305
 26 15 344,043 336,721 154,972
 27 16 331,704 320,505 121,316
 28 17 318,374 303,625 86,282
 29 18 303,971 286,054 49,814
 30 19 288,411 267,764 11,854
 31 20 271,600 248,725
 32 21 253,436 228,907
 33 22 233,813 208,277
 34 23 212,612 186,803
 35 24 189,706 164,450
 36 25 164,959 141,182
 37 26 138,222 116,961
 38 27 109,335 91,749
 39 28 78,125 65,505
 40 29 44,407 38,187
 41 30 7,977 9,750
* Substituting prices 10/27/2017 lowers the ladder's cost a touch from $804,000 to $800,000.
Last edited by #Cruncher on Sat Oct 28, 2017 5:52 pm, edited 1 time in total.

User avatar
Phineas J. Whoopee
Posts: 6656
Joined: Sun Dec 18, 2011 6:18 pm

Re: SWR with a real return of 1.6%

Post by Phineas J. Whoopee » Sat Oct 28, 2017 5:25 pm

Ron Scott wrote:
Sat Oct 28, 2017 8:01 am
... Safe and perpetual withdrawal rate calculators (like this one: https://portfoliocharts.com/portfolio/withdrawal-rates/) show that taking some risk in stock/bond allocations increases the %s you can safely withdraw, i.e., the more you earn on the nest egg the more you can spend.
...
The study that concluded 4% of initial value, adjusted each year for CPI, was sustainable didn't mention anything at all about being safe, and very particularly was not about being perpetual. It was published at a time when many financial advisers were suggesting 8% and more was sustainable. At least one famous radio personality today says 12% will reliably work.

Four percent was the highest level, adjusted each year for inflation, at which the model portfolio did not fall to zero prior to thirty years in the worst of the historical periods examined. Portfolios with withdrawal rates above 4% very much did fall to zero in many scenarios. Even 4% fell to zero remaining portfolio value after thirty years in the most unfavorable period the authors looked at.

I don't know where the word safe came from, but it's a mighty matter of legend.

If you want safe you do not want the Trinity Study. If you want perpetual you do not want the Trinity Study. If you want a rational plan for drawing from your portfolio over many years you do not want the Trinity Study.

If you want a rough guideline about whether you have enough accumulated to retire at a normal age and cover your remaining costs after any pensions and Social Security the Trinity Study can be useful.

PJW

dbr
Posts: 23687
Joined: Sun Mar 04, 2007 9:50 am

Re: SWR with a real return of 1.6%

Post by dbr » Sun Oct 29, 2017 9:39 am

#Cruncher wrote:
Sat Oct 28, 2017 10:56 am
TIPS yields aren't quite that high now, dbr. Better to have said:
dbr wrote:
Sat Oct 28, 2017 8:40 am
You can [If you could] set up a ladder of 30 year TIPS at 1.3% real and [you'd] get 4% real back for 30 years.
But your 4% is close anyway. According to my latest TIPS Ladder Builder spreadsheet, with 07/24/2017 prices, $804,000 [*] buys enough TIPS to provide $30,000 in constant dollars for 30 years, a "withdrawal rate" of 3.7%.
Oh, you mean setting up an LMP does not immunize you to the cruel fate of history being what it is? :shock:

User avatar
siamond
Posts: 3426
Joined: Mon May 28, 2012 5:50 am

Re: SWR with a real return of 1.6%

Post by siamond » Sun Oct 29, 2017 10:31 am

Ron Scott wrote:
Fri Oct 27, 2017 7:44 pm
I'm inclined to base my financial decisions for the foreseeable future on Bogle's most recent assumptions about future-decade returns.
Jack Bogle's expected returns is for the mid-term (e.g. coming decade), assuming a return of valuations to some sort of normal for stocks, and a return of yield to some sort of normal for bonds. Remember, this is probabilistic, the center point of a wide cloud of possible outcomes.

The financial decisions you are speaking of (e.g. SWR) should be about your entire retirement period, which I hope will be significantly more than a decade.

If & when valuations come down to Earth, say in a decade, the period AFTER that would display more regular expected returns. If you follow (and accept) John's math in his slideset, the real expected returns for stocks after such first troubled decade would be back to 4% or so for stocks. Similarly, bonds returns would be in better shape once yields get more average.

There is NO reason to change your SWR based on such dire expected returns for the coming decade. The SWR math *already* includes historical situations which were as dire, if not more. Then things came back to some sort of normal, and life went on.

TheNightsToCome
Posts: 125
Joined: Fri Jun 30, 2017 11:48 pm

Re: SWR with a real return of 1.6%

Post by TheNightsToCome » Sun Oct 29, 2017 11:03 am

siamond wrote:
Sun Oct 29, 2017 10:31 am
Ron Scott wrote:
Fri Oct 27, 2017 7:44 pm
I'm inclined to base my financial decisions for the foreseeable future on Bogle's most recent assumptions about future-decade returns.
Jack Bogle's expected returns is for the mid-term (e.g. coming decade), assuming a return of valuations to some sort of normal for stocks, and a return of yield to some sort of normal for bonds. Remember, this is probabilistic, the center point of a wide cloud of possible outcomes.

The financial decisions you are speaking of (e.g. SWR) should be about your entire retirement period, which I hope will be significantly more than a decade.

If & when valuations come down to Earth, say in a decade, the period AFTER that would display more regular expected returns. If you follow (and accept) John's math in his slideset, the real expected returns for stocks after such first troubled decade would be back to 4% or so for stocks. Similarly, bonds returns would be in better shape once yields get more average.

There is NO reason to change your SWR based on such dire expected returns for the coming decade. The SWR math *already* includes historical situations which were as dire, if not more. Then things came back to some sort of normal, and life went on.
"The SWR math *already* includes historical situations which were as dire, if not more. Then things came back to some sort of normal, and life went on."

I'm not sure things have ever been this "dire," given the current combination of extremely high valuations for both stocks and bonds. (LT TIPS provided a 4-4.5% real yield in 2000 when stock valuations were at record highs.) Even if so, that "n" is very small.

Future returns decline and the risk of large draw-downs increases as valuations climb. Bengen's original paper and the Trinity study used the entire historical data set from time A to time B. However, the data set relevant to current conditions is not the universe of all starting dates, but only those with similar valuations. When we include starting valuations that are below average and average, we inflate the success rate of a given withdrawal rate.

Future success rates for 4% starting today are likely to be much lower than in the past (and are also lower using data from 1900-2014 in the US, and much lower with that data set in other developed nations than in the US).

dbr
Posts: 23687
Joined: Sun Mar 04, 2007 9:50 am

Re: SWR with a real return of 1.6%

Post by dbr » Sun Oct 29, 2017 11:14 am

TheNightsToCome wrote:
Sun Oct 29, 2017 11:03 am

Future returns decline and the risk of large draw-downs increases as valuations climb. Bengen's original paper and the Trinity study used the entire historical data set from time A to time B. However, the data set relevant to current conditions is not the universe of all starting dates, but only those with similar valuations. When we include starting valuations that are below average and average, we inflate the success rate of a given withdrawal rate.

Future success rates for 4% starting today are likely to be much lower than in the past (and are also lower using data from 1900-2014 in the US, and much lower with that data set in other developed nations than in the US).
Your observation is correct, but . . .

An attempt can be made to compute portfolio failure probabilities conditional on knowing something of current conditions. It is already tough enough to get unconditional results for this stuff. Getting good numbers for a subset of that universe is not easy. Note that what is required is a statistical projection of the effect of current conditions on an entire retirement, and that is not easy to come by compared to mooting conclusion about the next ten years or something. A lot of water has yet to pass under the bridge.

A problem seemingly shared by all SWR studies is that those probabilities come without estimates of uncertainty attached. I think the methodology required to assess that is a sensitivity study of how variation in input assumptions, such as expected return, affect the result. We know withdrawal rate has a large effect and asset allocation does not have a large effect. Then you need a method to come up with the 30 year future estimate of expected return if you are going to vary that.

User avatar
siamond
Posts: 3426
Joined: Mon May 28, 2012 5:50 am

Re: SWR with a real return of 1.6%

Post by siamond » Sun Oct 29, 2017 12:36 pm

TheNightsToCome wrote:
Sun Oct 29, 2017 11:03 am
I'm not sure things have ever been this "dire," given the current combination of extremely high valuations for both stocks and bonds. (LT TIPS provided a 4-4.5% real yield in 2000 when stock valuations were at record highs.) Even if so, that "n" is very small. [...]

When we include starting valuations that are below average and average, we inflate the success rate of a given withdrawal rate.
Yes, those are fair points. But the OP wasn't asking about success rates. And his math just doesn't add up, for the reasons I identified, which was the primary point I was trying to convey. The SWR doesn't change just because you may be at a high valuation point.

This being said, I would tend to agree that future returns might become more muted than they were in a fairly rosy past (at least in the US), and that, in addition, the current combination is indeed concerning. This has been discussed ad nauseam on this forum, and some folks reach the conclusion that the 'new SWR' might be 3.5% (which isn't crazy considering that other countries did experience such painful numbers in their past). But certainly not 1.6%!!

TheNightsToCome
Posts: 125
Joined: Fri Jun 30, 2017 11:48 pm

Re: SWR with a real return of 1.6%

Post by TheNightsToCome » Sun Oct 29, 2017 1:11 pm

siamond wrote:
Sun Oct 29, 2017 12:36 pm
TheNightsToCome wrote:
Sun Oct 29, 2017 11:03 am
I'm not sure things have ever been this "dire," given the current combination of extremely high valuations for both stocks and bonds. (LT TIPS provided a 4-4.5% real yield in 2000 when stock valuations were at record highs.) Even if so, that "n" is very small. [...]
When we include starting valuations that are below average and average, we inflate the success rate of a given withdrawal rate.
Yes, those are fair points. But the OP wasn't asking about success rates. And his math just doesn't add up, for the reasons I identified, which was the primary point I was trying to convey. The SWR doesn't change just because you may be at a high valuation point.

This being said, I would tend to agree that future returns might become more muted than they were in a fairly rosy past (at least in the US), and that, in addition, the current combination is indeed concerning. This has been discussed ad nauseam on this forum, and some folks reach the conclusion that the 'new SWR' might be 3.5% (which isn't crazy considering that other countries did experience such painful numbers in their past). But certainly not 1.6%!!
If I read Table 1 correctly (see link below to Wade Pfau study), then the SWR between 1900-2012 was less than 1.6% for seven out of 20 developed nations (and no greater than 3.68% in any nation):

https://advisorperspectives.com/art ... awal-rates

Valuations have probably never been less favorable than today, so it isn't difficult for me to imagine that the SWR will turn out to be 1.6% over the next 30 years.

Of course, those results were with 50:50 stock:bond portfolios subject to sequence-of-returns-risk. If a retiree can set up a liability matching portfolio with a guaranteed zero percent after-tax real return, then he/she can spend 3.3% of a portfolio for 30 years before running out of money. That is a higher SWR than 15/20 developed nations between 1900-2012, and only 0.37% below the SWR in the US during that period.

randomguy
Posts: 4871
Joined: Wed Sep 17, 2014 9:00 am

Re: SWR with a real return of 1.6%

Post by randomguy » Sun Oct 29, 2017 1:30 pm

TheNightsToCome wrote:
Sun Oct 29, 2017 1:11 pm
siamond wrote:
Sun Oct 29, 2017 12:36 pm
TheNightsToCome wrote:
Sun Oct 29, 2017 11:03 am
I'm not sure things have ever been this "dire," given the current combination of extremely high valuations for both stocks and bonds. (LT TIPS provided a 4-4.5% real yield in 2000 when stock valuations were at record highs.) Even if so, that "n" is very small. [...]
When we include starting valuations that are below average and average, we inflate the success rate of a given withdrawal rate.
Yes, those are fair points. But the OP wasn't asking about success rates. And his math just doesn't add up, for the reasons I identified, which was the primary point I was trying to convey. The SWR doesn't change just because you may be at a high valuation point.

This being said, I would tend to agree that future returns might become more muted than they were in a fairly rosy past (at least in the US), and that, in addition, the current combination is indeed concerning. This has been discussed ad nauseam on this forum, and some folks reach the conclusion that the 'new SWR' might be 3.5% (which isn't crazy considering that other countries did experience such painful numbers in their past). But certainly not 1.6%!!
If I read Table 1 correctly (see link below to Wade Pfau study), then the SWR between 1900-2012 was less than 1.6% for seven out of 20 developed nations (and no greater than 3.68% in any nation):

https://advisorperspectives.com/art ... awal-rates

Valuations have probably never been less favorable than today, so it isn't difficult for me to imagine that the SWR will turn out to be 1.6% over the next 30 years.

Of course, those results were with 50:50 stock:bond portfolios subject to sequence-of-returns-risk. If a retiree can set up a liability matching portfolio with a guaranteed zero percent after-tax real return, then he/she can spend 3.3% of a portfolio for 30 years before running out of money. That is a higher SWR than 15/20 developed nations between 1900-2012, and only 0.37% below the SWR in the US during that period.
And how many of those countries were leveled 2x or had a political revolution? Your LMP might not help much in those cases. Even international diversification will not help if you can't get access to your money

User avatar
siamond
Posts: 3426
Joined: Mon May 28, 2012 5:50 am

Re: SWR with a real return of 1.6%

Post by siamond » Sun Oct 29, 2017 1:38 pm

TheNightsToCome wrote:
Sun Oct 29, 2017 1:11 pm
If I read Table 1 correctly (see link below to Wade Pfau study), then the SWR between 1900-2012 was less than 1.6% for seven out of 20 developed nations (and no greater than 3.68% in any nation):

https://advisorperspectives.com/art ... awal-rates

Valuations have probably never been less favorable than today, so it isn't difficult for me to imagine that the SWR will turn out to be 1.6% over the next 30 years.
Take a closer look at those seven countries... Why did they fail so spectacularly? Because of two domestically devastating world wars after the 1900-1912 starting point. Nothing to do with high valuations. Everything to do with unpredictable black swans. Which may occur in the US, mind you, but the linkage you're making with valuations is nowhere near as meaningful as you seem to make it.

Look, you are absolutely right that current valuations ARE (very) concerning in the US, no discussion about that, but NOT to the point of inferring that the SWR will shed 2.5 points overnight.

TheNightsToCome
Posts: 125
Joined: Fri Jun 30, 2017 11:48 pm

Re: SWR with a real return of 1.6%

Post by TheNightsToCome » Sun Oct 29, 2017 2:10 pm

siamond wrote:
Sun Oct 29, 2017 1:38 pm
TheNightsToCome wrote:
Sun Oct 29, 2017 1:11 pm
If I read Table 1 correctly (see link below to Wade Pfau study), then the SWR between 1900-2012 was less than 1.6% for seven out of 20 developed nations (and no greater than 3.68% in any nation):

https://advisorperspectives.com/art ... awal-rates

Valuations have probably never been less favorable than today, so it isn't difficult for me to imagine that the SWR will turn out to be 1.6% over the next 30 years.
Take a closer look at those seven countries... Why did they fail so spectacularly? Because of two domestically devastating world wars after the 1900-1912 period. Nothing to do with high valuations. Everything to do with unpredictable black swans. Which may occur in the US, mind you, but the linkage you're making with valuations is nowhere near as meaningful as you seem to make it. Current valuations ARE concerning, no discussion about that, but NOT to the point of inferring that the SWR will shed 2.5 points overnight.
"but the linkage you're making with valuations is nowhere near as meaningful as you seem to make it."

That's probably fair. That is, I presume the lowest SAFEMAX values in those countries were due to wars, although it seems likely that Japan's SAFEMAX since 1989 will turn out to be extremely low, and that is related to high valuations rather than war.

Still, only 5/20 nations had a SAFEMAX higher than the 3.33% one can obtain with a LMP, and none of them significantly higher. (The 4% rule failed 8.3% of the time in the US.) I think the OP is correct to be uncomfortable with the 4% rule (which has an expected failure probability much higher than 8.3% today).

If he has a standard 30-year horizon, I'd recommend a LMP with a withdrawal rate of about 3.33%. If he plans an early, long retirement then he may have to take more risk, and a 3% withdrawal rate will probably work, but I wouldn't consider it a conservative, forget-all-worries safe plan for 50 year horizons.

randomguy
Posts: 4871
Joined: Wed Sep 17, 2014 9:00 am

Re: SWR with a real return of 1.6%

Post by randomguy » Sun Oct 29, 2017 2:48 pm

TheNightsToCome wrote:
Sun Oct 29, 2017 2:10 pm
siamond wrote:
Sun Oct 29, 2017 1:38 pm
TheNightsToCome wrote:
Sun Oct 29, 2017 1:11 pm
If I read Table 1 correctly (see link below to Wade Pfau study), then the SWR between 1900-2012 was less than 1.6% for seven out of 20 developed nations (and no greater than 3.68% in any nation):

https://advisorperspectives.com/art ... awal-rates

Valuations have probably never been less favorable than today, so it isn't difficult for me to imagine that the SWR will turn out to be 1.6% over the next 30 years.
Take a closer look at those seven countries... Why did they fail so spectacularly? Because of two domestically devastating world wars after the 1900-1912 period. Nothing to do with high valuations. Everything to do with unpredictable black swans. Which may occur in the US, mind you, but the linkage you're making with valuations is nowhere near as meaningful as you seem to make it. Current valuations ARE concerning, no discussion about that, but NOT to the point of inferring that the SWR will shed 2.5 points overnight.
"but the linkage you're making with valuations is nowhere near as meaningful as you seem to make it."

That's probably fair. That is, I presume the lowest SAFEMAX values in those countries were due to wars, although it seems likely that Japan's SAFEMAX since 1989 will turn out to be extremely low, and that is related to high valuations rather than war.

Still, only 5/20 nations had a SAFEMAX higher than the 3.33% one can obtain with a LMP, and none of them significantly higher. (The 4% rule failed 8.3% of the time in the US.) I think the OP is correct to be uncomfortable with the 4% rule (which has an expected failure probability much higher than 8.3% today).

If he has a standard 30-year horizon, I'd recommend a LMP with a withdrawal rate of about 3.33%. If he plans an early, long retirement then he may have to take more risk, and a 3% withdrawal rate will probably work, but I wouldn't consider it a conservative, forget-all-worries safe plan for 50 year horizons.
What was the SWR that a LMP would have given in any of those countries? Your assumption that getting invaded/political revolution wouldn't affect your LMP is optimistic at best.

User avatar
siamond
Posts: 3426
Joined: Mon May 28, 2012 5:50 am

Re: SWR with a real return of 1.6%

Post by siamond » Sun Oct 29, 2017 2:54 pm

TheNightsToCome wrote:
Sun Oct 29, 2017 2:10 pm
"but the linkage you're making with valuations is nowhere near as meaningful as you seem to make it."

That's probably fair. That is, I presume the lowest SAFEMAX values in those countries were due to wars, although it seems likely that Japan's SAFEMAX since 1989 will turn out to be extremely low, and that is related to high valuations rather than war.
Definitely true about Japan, although the US is nowhere near Japan's sky-high valuation peak. I actually did the SWR math for the modern times (1970+) for 16 countries (incl. Japan) in this 3-parts blog article. Your assumption about Japan is absolutely correct IF the investor would have stayed 100% domestic. A more international mix would definitely have helped to restore an SWR to 3.5% or so. And that's a message I believe US investors should heed...
TheNightsToCome wrote:
Sun Oct 29, 2017 2:10 pm
Still, only 5/20 nations had a SAFEMAX higher than the 3.33% one can obtain with a LMP, and none of them significantly higher. (The 4% rule failed 8.3% of the time in the US.) I think the OP is correct to be uncomfortable with the 4% rule (which has an expected failure probability much higher than 8.3% today).
Oh, I am quite concerned too, don't misread me. I was just reacting to the incorrect math mixing up 10-years expected returns with SWR numbers, and ending up with 1.6%. But the concern definitely remains, agreed. Mitigation does seem in order, and it could be a mix of LMP, of hedging domestic/international bets, of using a variable withdrawal method (instead of the brain-dead fixed withdrawal associated with the SWR math). Personally, the LMP concept leaves me cold and I concentrate on the two others, but there are many roads to financial safety (I hope!). Good discussion.

TheNightsToCome
Posts: 125
Joined: Fri Jun 30, 2017 11:48 pm

Re: SWR with a real return of 1.6%

Post by TheNightsToCome » Sun Oct 29, 2017 4:31 pm

randomguy wrote:
Sun Oct 29, 2017 2:48 pm
TheNightsToCome wrote:
Sun Oct 29, 2017 2:10 pm
siamond wrote:
Sun Oct 29, 2017 1:38 pm
TheNightsToCome wrote:
Sun Oct 29, 2017 1:11 pm
If I read Table 1 correctly (see link below to Wade Pfau study), then the SWR between 1900-2012 was less than 1.6% for seven out of 20 developed nations (and no greater than 3.68% in any nation):

https://advisorperspectives.com/art ... awal-rates

Valuations have probably never been less favorable than today, so it isn't difficult for me to imagine that the SWR will turn out to be 1.6% over the next 30 years.
Take a closer look at those seven countries... Why did they fail so spectacularly? Because of two domestically devastating world wars after the 1900-1912 period. Nothing to do with high valuations. Everything to do with unpredictable black swans. Which may occur in the US, mind you, but the linkage you're making with valuations is nowhere near as meaningful as you seem to make it. Current valuations ARE concerning, no discussion about that, but NOT to the point of inferring that the SWR will shed 2.5 points overnight.
"but the linkage you're making with valuations is nowhere near as meaningful as you seem to make it."

That's probably fair. That is, I presume the lowest SAFEMAX values in those countries were due to wars, although it seems likely that Japan's SAFEMAX since 1989 will turn out to be extremely low, and that is related to high valuations rather than war.

Still, only 5/20 nations had a SAFEMAX higher than the 3.33% one can obtain with a LMP, and none of them significantly higher. (The 4% rule failed 8.3% of the time in the US.) I think the OP is correct to be uncomfortable with the 4% rule (which has an expected failure probability much higher than 8.3% today).

If he has a standard 30-year horizon, I'd recommend a LMP with a withdrawal rate of about 3.33%. If he plans an early, long retirement then he may have to take more risk, and a 3% withdrawal rate will probably work, but I wouldn't consider it a conservative, forget-all-worries safe plan for 50 year horizons.
What was the SWR that a LMP would have given in any of those countries? Your assumption that getting invaded/political revolution wouldn't affect your LMP is optimistic at best.
Yes, if the government is toppled and my LMP is in that government's TIPS I'll probably be out of luck. For that matter, if the government is bankrupt (e.g., Argentina), I'll be out of luck.

The SWR for the bottom countries were:

Japan 0.25%
Austria 0.26%
Germany 0.84%
Italy 0.9%

Are the SWRs low for these countries because they defaulted on bonds? I don't know.

France 0.93%
Finland 1.34%
Belgium 1.39%

I don't know the history for this group of three nations with SWR below 1.6%, but I don't believe that the governments were overthrown or that any defaulted on their sovereign bonds. Is that wrong?

Their SWRs might be this low because of stock market losses due to war. I don't know that for sure either.

Spain 2.18%
Ireland 2.6%
Norway 2.72%
Australia 2.9%
Switzerland 2.92%
UK 3.05%
Sweden 3.23%
Netherlands 3.24%

The fact that this group had SWRs below that available on a LMP (3.33%) is shocking.

South Africa 3.46%
New Zealand 3.6%
Canada 3.66%
US 3.67% (not 4%)
Denmark 3.68%

With a 50:50 portfolio, your champion (Denmark) provided an extra 0.35% in SWR above that obtained by a LMP, but 3/4 of the nations did worse. The LMP might not have saved a retiree if the government was destroyed, but that retiree would also be ruined if he held a 50:50 portfolio.

The bottom line is that we don't know what the prospective SWR is today for a 50:50 portfolio, but it may well be less than 3%. We know for sure that 3.33% is safe with a LMP--for all outcomes without a US default. You're taking a bigger risk using a withdrawal rate above 3.33% and a stock/bond portfolio.

TheNightsToCome
Posts: 125
Joined: Fri Jun 30, 2017 11:48 pm

Re: SWR with a real return of 1.6%

Post by TheNightsToCome » Sun Oct 29, 2017 4:45 pm

siamond wrote:
Sun Oct 29, 2017 2:54 pm
TheNightsToCome wrote:
Sun Oct 29, 2017 2:10 pm
"but the linkage you're making with valuations is nowhere near as meaningful as you seem to make it."

That's probably fair. That is, I presume the lowest SAFEMAX values in those countries were due to wars, although it seems likely that Japan's SAFEMAX since 1989 will turn out to be extremely low, and that is related to high valuations rather than war.
Definitely true about Japan, although the US is nowhere near Japan's sky-high valuation peak. I actually did the SWR math for the modern times (1970+) for 16 countries (incl. Japan) in this 3-parts blog article. Your assumption about Japan is absolutely correct IF the investor would have stayed 100% domestic. A more international mix would definitely have helped to restore an SWR to 3.5% or so. And that's a message I believe US investors should heed...
TheNightsToCome wrote:
Sun Oct 29, 2017 2:10 pm
Still, only 5/20 nations had a SAFEMAX higher than the 3.33% one can obtain with a LMP, and none of them significantly higher. (The 4% rule failed 8.3% of the time in the US.) I think the OP is correct to be uncomfortable with the 4% rule (which has an expected failure probability much higher than 8.3% today).
Oh, I am quite concerned too, don't misread me. I was just reacting to the incorrect math mixing up 10-years expected returns with SWR numbers, and ending up with 1.6%. But the concern definitely remains, agreed. Mitigation does seem in order, and it could be a mix of LMP, of hedging domestic/international bets, of using a variable withdrawal method (instead of the brain-dead fixed withdrawal associated with the SWR math). Personally, the LMP concept leaves me cold and I concentrate on the two others, but there are many roads to financial safety (I hope!). Good discussion.
"A more international mix would definitely have helped to restore an SWR to 3.5% or so. And that's a message I believe US investors should heed..."

I haven't read your blog but I'll put it on my reading list.

Pfau's study (https://advisorperspectives.com/art ... awal-rates) appears to show that a 3.5% "SWR" with a 50:50 global stock:bond portfolio worked in only 3/20 nations.

Valuations are very high around the world now, not just in US. Why would you be comfortable with 3.5% in a global portfolio?

"Personally, the LMP concept leaves me cold and I concentrate on the two others, but there are many roads to financial safety (I hope!)."

It also leaves me "cold," but that is because I hope to become an active investor after I retire. (I'm a former healthcare equity analyst.)

Most SatuMedia are conservative (and obviously don't hope to beat the market as active investors), so I'm surprised that liability matching isn't a more popular choice on this forum.

User avatar
siamond
Posts: 3426
Joined: Mon May 28, 2012 5:50 am

Re: SWR with a real return of 1.6%

Post by siamond » Sun Oct 29, 2017 5:32 pm

TheNightsToCome wrote:
Sun Oct 29, 2017 4:45 pm
Pfau's study (https://advisorperspectives.com/art ... awal-rates) appears to show that a 3.5% "SWR" with a 50:50 global stock:bond portfolio worked in only 3/20 nations.
Thanks, I hadn't read this one. Just skimmed through it, it's a bit hard to read with the poorly formatted tables. And I don't have access to the DMS dataset ($$$!), so this is something I can't reproduce. Can't find the article on SSRN. Will ask Wade for his original. His top-level conclusion lines up with mine in terms of hedging bets, but yes, the absolute numbers are frighteningly low.
TheNightsToCome wrote:
Sun Oct 29, 2017 4:45 pm
Valuations are very high around the world now, not just in US. Why would you be comfortable with 3.5% in a global portfolio?
As far as I know, the best reference for US vs. Int'l valuations is Star Capital. Showing that the US seems to be in bleeding territory, while the rest of the world is not as hot, far from it. Still, the RoW does appear high, granted. I still don't see that this creates a totally unprecedented situation, but it is hard to say as CAPE historical stats are sorely lacking outside the US.

I never said I am comfortable with a 3.5% SWR. I just observed that in since 1970, it would appear to have usually worked ok *IF* solid international diversification had been used. Now in time of domestically-impacting war(s), all bets are off... I would still want to be internationally diversified in such a case, mind you. All this being said, the sheer concept of 'SWR' makes limited sense. It's basically a single data point in past history, and based on a very silly approach to withdrawals. It is also non-sensical to gate everything in life by the worst possible case. So personally, I don't obsess over it. But I keep agreeing with you that the current situation is quite concerning. :wink:

AlohaJoe
Posts: 2550
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: SWR with a real return of 1.6%

Post by AlohaJoe » Sun Oct 29, 2017 7:59 pm

Just a note that Pfau's SWR numbers are based on DMS bonds data. That data is about long bonds. Those with 30 year terms. The ones that SatuMedia generally believe are too risky to include in any portfolio. Pfau (and DMS) use them because intermediate bonds were not common across governments until fairly recently. It is well known that moving away from intermediate government bonds to things with more risk lowers the withdrawal rate with US data so I would assume that is a general trend.

TheNightsToCome
Posts: 125
Joined: Fri Jun 30, 2017 11:48 pm

Re: SWR with a real return of 1.6%

Post by TheNightsToCome » Sun Oct 29, 2017 8:22 pm

AlohaJoe wrote:
Sun Oct 29, 2017 7:59 pm
Just a note that Pfau's SWR numbers are based on DMS bonds data. That data is about long bonds. Those with 30 year terms. The ones that SatuMedia generally believe are too risky to include in any portfolio. Pfau (and DMS) use them because intermediate bonds were not common across governments until fairly recently. It is well known that moving away from intermediate government bonds to things with more risk lowers the withdrawal rate with US data so I would assume that is a general trend.
The article (https://advisorperspectives.com/art ... awal-rates) doesn't provide any details about the bond portion of the portfolio, and Dimson's Triumph of the Optimists provides data on both "bills" and "bonds."

How did you find more details about Pfau's bond data for this study? Do you have a link to a full paper (hopefully with clear formatting of tables)?

Much appreciated if you do.

AlohaJoe
Posts: 2550
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: SWR with a real return of 1.6%

Post by AlohaJoe » Sun Oct 29, 2017 8:41 pm

From Pfau's recent book, when taking about global SWR he writes
Bengen used intermediate-term government bonds—which seem to hold a sweet spot in the trade-off between returns and volatility—to yield the highest sustainable spending. But this bond option is not available in the Global Returns Dataset. For what is available, bills work better than long-term bonds for the aforementioned reason that their lower volatility helps more than the lower return hurts.

AlohaJoe
Posts: 2550
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: SWR with a real return of 1.6%

Post by AlohaJoe » Sun Oct 29, 2017 8:53 pm

I should also add that I believe Pfau's numbers are generally based on a 100% stock portfolio because using bonds at all in the 20th century was usually catastrophic. Curiously, that part of Pfau's findings is usually ignored (even by Pfau).

While lessons can be drawn from global historical data, it is hard to know what those lessons should be :). It is a bit like a rorshach test. For the average conservative Boglehead, they probably see it as evidence that withdrawals rates should be lower. Others, like me, see evidence that bonds shouldn't be used trusted :).
Or maybe the lesson is "don't have a world war kill millions within your borders"? Or maybe it is "inflation is the ultimate evil but The Great Moderation has killed it"? Or maybe it is that "weak central banks are the fundamental cause"? Or....I dunno. It is all pretty complicated!

TheNightsToCome
Posts: 125
Joined: Fri Jun 30, 2017 11:48 pm

Re: SWR with a real return of 1.6%

Post by TheNightsToCome » Sun Oct 29, 2017 8:53 pm

AlohaJoe wrote:
Sun Oct 29, 2017 8:41 pm
From Pfau's recent book, when taking about global SWR he writes
Bengen used intermediate-term government bonds—which seem to hold a sweet spot in the trade-off between returns and volatility—to yield the highest sustainable spending. But this bond option is not available in the Global Returns Dataset. For what is available, bills work better than long-term bonds for the aforementioned reason that their lower volatility helps more than the lower return hurts.
Well, based on this it seems likely that Pfau would choose the bills data for his SWR study. Why wouldn't he? Dimson et al have data on bills. Otherwise, he wouldn't obtain the best SWR.

TheNightsToCome
Posts: 125
Joined: Fri Jun 30, 2017 11:48 pm

Re: SWR with a real return of 1.6%

Post by TheNightsToCome » Sun Oct 29, 2017 8:55 pm

AlohaJoe wrote:
Sun Oct 29, 2017 8:53 pm
I should also add that I believe Pfau's numbers are generally based on a 100% stock portfolio because using bonds at all in the 20th century was usually catastrophic. Curiously, that part of Pfau's findings is usually ignored (even by Pfau).

While lessons can be drawn from global historical data, it is hard to know what those lessons should be :). It is a bit like a rorshach test. For the average conservative Boglehead, they probably see it as evidence that withdrawals rates should be lower. Others, like me, see evidence that bonds shouldn't be used trusted :).
Or maybe the lesson is "don't have a world war kill millions within your borders"? Or maybe it is "inflation is the ultimate evil but The Great Moderation has killed it"? Or maybe it is that "weak central banks are the fundamental cause"? Or....I dunno. It is all pretty complicated!
"I should also add that I believe Pfau's numbers are generally based on a 100% stock portfolio"

Well, this is definitely not the case. In the study I linked to above, he clearly states that the portfolios were 50% fixed income and 50% stocks.

AlohaJoe
Posts: 2550
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: SWR with a real return of 1.6%

Post by AlohaJoe » Sun Oct 29, 2017 9:09 pm

Pfau has written a few articles with variants on this and I admit that I didn't click to read the one you linked. Of his original article on global SWRs Pfau writes
In the original article, to avoid claims of bias that I chose an asset allocation that exaggerated the risk of the 4 percent rule, I used a rather generous and unrealistic “perfect foresight assumption.” For each country and in each retirement year, I used the asset allocation for local market investors (between stocks, bonds, and bills) that supported the highest withdrawal rate. In many cases, though not all, this meant using 100 percent stocks.

Ron Scott
Posts: 339
Joined: Tue Apr 05, 2016 5:38 am

Re: SWR with a real return of 1.6%

Post by Ron Scott » Sun Oct 29, 2017 9:11 pm

Fascinating discussion and thanks to the contributors.

For those of you who have spent time with the research in this area two questions:

1. Are you concerned that the data available for consideration is historic and that the models can't account for any sea-change in economics, especially US economics?

2. Let's double down, and assume that past results are a good indicator of future performance: What percent of the variance in the projected future years performance in these models is actually explained by the historical data?

TheNightsToCome
Posts: 125
Joined: Fri Jun 30, 2017 11:48 pm

Re: SWR with a real return of 1.6%

Post by TheNightsToCome » Sun Oct 29, 2017 9:40 pm

#Cruncher wrote:
Sat Oct 28, 2017 10:56 am
TIPS yields aren't quite that high now, dbr. Better to have said:
dbr wrote:
Sat Oct 28, 2017 8:40 am
You can [If you could] set up a ladder of 30 year TIPS at 1.3% real and [you'd] get 4% real back for 30 years.
But your 4% is close anyway. According to my latest TIPS Ladder Builder spreadsheet, with 07/24/2017 prices, $804,000 [*] buys enough TIPS to provide $30,000 in constant dollars for 30 years, a "withdrawal rate" of 3.7%.
Ron Scott wrote:
Sat Oct 28, 2017 9:41 am
Interesting. Then what's the point of taking risk in retirement with stocks, etc.
Assuming you're referring to dbr's TIPS ladder, the reason one might take risk is the potential to have money left over after 30 years. There is no chance of this with the TIPS ladder.

For what it's worth, here is my estimate that if a 4.8% return supports a 4% withdrawal rate, then a 1.6% return would support a 3% withdrawal. This is based on the severe case where the portfolio falls 10% every year for the first five years and then grows at a constant rate for the next 25 years such that the overall 30 year growth rate is 4.8% or 1.6%. Given this assumption, both the 4.8% growth/4% withdrawal and 1.6% growth/3% withdrawal scenarios last 30 years before failing. However, a 4% withdrawal with only 1.6% overall growth would fail after 19 years. (For more on this simplistic methodology, see my post, Re: Avoiding the Spiral.)

Code: Select all

 1 Initial balance 1,000,000 1,000,000 1,000,000
 2 Withdrawal 4.000% 3.000% 4.000% <--
 5 Portfolio average growth 4.800% 1.600% 1.600% <--
 6 Over this many years 30 30 30
 7 But for first [Y] years 5 5 5
 8 It falls [F] (10.000%) (10.000%) (10.000%)
 9 Thereafter it rises [R] 8.040% 4.094% 4.094%
 Year ------------ Balance ------------

Code: Select all

 11 0 1,000,000 1,000,000 1,000,000
 12 1 860,000 870,000 860,000
 13 2 734,000 753,000 734,000
 14 3 620,600 647,700 620,600
 15 4 518,540 552,930 518,540
 16 5 426,686 467,637 426,686
 17 6 420,992 456,780 404,153
 18 7 414,840 445,479 380,697
 19 8 408,194 433,715 356,281
 20 9 401,013 421,469 330,866
 21 10 393,255 408,722 304,410
 22 11 384,873 395,454 276,871
 23 12 375,818 381,642 248,205
 24 13 366,034 367,265 218,366
 25 14 355,463 352,299 187,305
 26 15 344,043 336,721 154,972
 27 16 331,704 320,505 121,316
 28 17 318,374 303,625 86,282
 29 18 303,971 286,054 49,814
 30 19 288,411 267,764 11,854
 31 20 271,600 248,725
 32 21 253,436 228,907
 33 22 233,813 208,277
 34 23 212,612 186,803
 35 24 189,706 164,450
 36 25 164,959 141,182
 37 26 138,222 116,961
 38 27 109,335 91,749
 39 28 78,125 65,505
 40 29 44,407 38,187
 41 30 7,977 9,750
* Substituting prices 10/27/2017 lowers the ladder's cost a touch from $804,000 to $800,000.
I hadn't read #Cruncher's informative post when I replied earlier about a 3.33% "withdrawal rate" using a LMP (composed of TIPS). I presumed that one could obtain at least 0% real after-tax, but I had not looked at actual numbers.

Based on #Cruncher's work, one can obtain a "withdrawal rate" greater than 3.7% using a LMP based on TIPS. That (3.7% SWR) is higher than any 30-year SWR in any nation using a domestic 50:50 portfolio, and it is higher than any SWR for any nation using a global 50:50 portfolio--except New Zealand (so it's better than 19 out of 20 nations, including the US).

If you are old enough (or sick enough) to be confident of dying within 30 years, this option looks attractive. Unlike an annuity, it preserves liquidity. An annuity would still be an option later in retirement.

AlphaLess
Posts: 46
Joined: Fri Sep 29, 2017 11:38 pm

Re: SWR with a real return of 1.6%

Post by AlphaLess » Sun Oct 29, 2017 10:25 pm

AlohaJoe wrote:
Sat Oct 28, 2017 8:18 am
With 0 volatility a 4% SWR only requires around 1.3% real returns.

So, yes, it is all about volatility. That's why retirement researchers have made such a big deal about "sequence of returns" over the past 30 years.

The 4% SWR is based on a 30 year period (1966-1996) when real returns were way above 1.3% real. But the volatility, 3 crashes in 8 years, is what caused problems.
Upvote times 100.

I would suggest those who disagree (or are unconvinced) with these statements SPEND A LOT OF TIME reading / using calculators that show relationship between return, volatility, and safe / permanent withdrawal rates.
Last edited by AlphaLess on Sun Oct 29, 2017 10:30 pm, edited 1 time in total.

AlphaLess
Posts: 46
Joined: Fri Sep 29, 2017 11:38 pm

Re: SWR with a real return of 1.6%

Post by AlphaLess » Sun Oct 29, 2017 10:28 pm

Ron Scott wrote:
Sun Oct 29, 2017 9:11 pm
1. Are you concerned that the data available for consideration is historic and that the models can't account for any sea-change in economics, especially US economics?
I have significant concerns that:

(a) existing data is HIGHLY overfit in various models: in other words, future may not resemble the past AT all, even if things such as CAGRs in the future are drawn from the same distribution as the past,

(b) various out-of-model events will result in future outcomes that are entirely not covered by models. E.g., nuclear war, global warming, geopolitical shifts that result in Western models of economy becoming inferior to others.

dbr
Posts: 23687
Joined: Sun Mar 04, 2007 9:50 am

Re: SWR with a real return of 1.6%

Post by dbr » Mon Oct 30, 2017 9:06 am

Ron Scott wrote:
Sun Oct 29, 2017 9:11 pm
Fascinating discussion and thanks to the contributors.

For those of you who have spent time with the research in this area two questions:

1. Are you concerned that the data available for consideration is historic and that the models can't account for any sea-change in economics, especially US economics?

Of course. But trying to overengineer retirement is silly anyway. If nothing else the uncertainties in non-financial life factors are huge and swamp out the uncertainties in financial factors. SWR studies are useful information about the nature of retirement finance but are not and were never intended as a plan to be followed with machine like precision. A good question to ask is how well anyone has experienced life up to now with everything neatly predicted according to design and then just extend that for the remainder.


2. Let's double down, and assume that past results are a good indicator of future performance: What percent of the variance in the projected future years performance in these models is actually explained by the historical data?

If the question for example, is what is the estimated error on the estimated SWR numbers, virtually no one who publishes this stuff attempts to make that estimate. If the statistics you are looking at are the final portfolio values as run out in FireCalc, then there is a distribution there that shows you the entire range of outcomes. You can calculate the average and standard deviation of that historical performance and estimate the future will look the same. Wade Pfau has run some Monte Carlo simulations where input data have been adjusted in accord with estimates of future conditions. You should probably look up his stuff.

A note is that if one wants to suspect that things are different this time it is much more likely to be wrong trying to guess why and how things are going to change than it is to guess things will be similar. Estimates of the future are strongly influenced by recency and political/emotional biases while history documents itself.



User avatar
siamond
Posts: 3426
Joined: Mon May 28, 2012 5:50 am

Re: SWR with a real return of 1.6%

Post by siamond » Mon Oct 30, 2017 2:16 pm

TheNightsToCome wrote:
Sun Oct 29, 2017 8:53 pm
AlohaJoe wrote:
Sun Oct 29, 2017 8:41 pm
From Pfau's recent book, when taking about global SWR he writes
Bengen used intermediate-term government bonds—which seem to hold a sweet spot in the trade-off between returns and volatility—to yield the highest sustainable spending. But this bond option is not available in the Global Returns Dataset. For what is available, bills work better than long-term bonds for the aforementioned reason that their lower volatility helps more than the lower return hurts.
Well, based on this it seems likely that Pfau would choose the bills data for his SWR study. Why wouldn't he? Dimson et al have data on bills. Otherwise, he wouldn't obtain the best SWR.
I swapped a couple of e-mails with Wade, and he clarified the following:

Note that this [DMS] dataset has long-term bonds and short-term bills. Short-term bills support higher withdrawal rates and so I usually use stocks/bills now. But this is an older article where I hadn’t yet settled on that convention. Withdrawal rates are a bit less because of the long-term bonds instead of bills.

Bengen's intermediate-term government bonds is the sweet spot for the highest withdrawal rates. For the Global returns data, it is long-term bonds or bills, and bills are the better choice of the two. The long-term bonds are too volatile and too exposed to inflation.


And here is a link to a clean copy of the corresponding article about International diversification and SWRs, with tables properly formatted. Again, this was done with LT bonds at the time of writing (2014).

User avatar
HomerJ
Posts: 10369
Joined: Fri Jun 06, 2008 12:50 pm

Re: SWR with a real return of 1.6%

Post by HomerJ » Mon Oct 30, 2017 3:46 pm

TheNightsToCome wrote:
Sun Oct 29, 2017 1:11 pm
siamond wrote:
Sun Oct 29, 2017 12:36 pm
TheNightsToCome wrote:
Sun Oct 29, 2017 11:03 am
I'm not sure things have ever been this "dire," given the current combination of extremely high valuations for both stocks and bonds. (LT TIPS provided a 4-4.5% real yield in 2000 when stock valuations were at record highs.) Even if so, that "n" is very small. [...]
When we include starting valuations that are below average and average, we inflate the success rate of a given withdrawal rate.
Yes, those are fair points. But the OP wasn't asking about success rates. And his math just doesn't add up, for the reasons I identified, which was the primary point I was trying to convey. The SWR doesn't change just because you may be at a high valuation point.

This being said, I would tend to agree that future returns might become more muted than they were in a fairly rosy past (at least in the US), and that, in addition, the current combination is indeed concerning. This has been discussed ad nauseam on this forum, and some folks reach the conclusion that the 'new SWR' might be 3.5% (which isn't crazy considering that other countries did experience such painful numbers in their past). But certainly not 1.6%!!
If I read Table 1 correctly (see link below to Wade Pfau study), then the SWR between 1900-2012 was less than 1.6% for seven out of 20 developed nations (and no greater than 3.68% in any nation):

https://advisorperspectives.com/art ... awal-rates

Valuations have probably never been less favorable than today, so it isn't difficult for me to imagine that the SWR will turn out to be 1.6% over the next 30 years.

Of course, those results were with 50:50 stock:bond portfolios subject to sequence-of-returns-risk. If a retiree can set up a liability matching portfolio with a guaranteed zero percent after-tax real return, then he/she can spend 3.3% of a portfolio for 30 years before running out of money. That is a higher SWR than 15/20 developed nations between 1900-2012, and only 0.37% below the SWR in the US during that period.
Wade Pfau also assumes 1% management fee, which completely skews his results compared to SatuMedia. Plus, he's been beating the doom drum since 2010, and we've gotten 12% real instead of the predicted 4% real. Yes, it's possible his models were correct, and we got really lucky... OR, maybe just maybe, its reasonable to wonder if maybe his models are missing a variable or two.

How long can PhDs make incorrect predictions before people stop listening to them? The truth is, economics really isn't a science. Too many variables (known and unknown), and the environment is constantly changing, with new laws, new industries, etc.

User avatar
siamond
Posts: 3426
Joined: Mon May 28, 2012 5:50 am

Re: SWR with a real return of 1.6%

Post by siamond » Mon Oct 30, 2017 4:14 pm

HomerJ wrote:
Mon Oct 30, 2017 3:46 pm
Wade Pfau also assumes 1% management fee, which completely skews his results compared to SatuMedia. Plus, he's been beating the doom drum since 2010, and we've gotten 12% real instead of the predicted 4% real. Yes, it's possible his models were correct, and we got really lucky... OR, maybe just maybe, its reasonable to wonder if maybe his models are missing a variable or two.
Maybe you should check facts a little bit before issuing such unwarranted criticism?

First off, the article we're discussing is about *past* Safe Withdrawal Rates (SWRs), nothing to do with your usual pet peeve about expected returns.

Next, this article is a continuation of a 2010 study about SWRs, where the author wrote the following:

4. Portfolio Administrative fees:
This assumption is rather vexing, but to be consistent with most studies […], we assume that mutual fund companies and financial advisers do not deduct any fees from the portfolio. We do this in order to make clear that the lower SAFEMAXs we find are due to factors other than such fees. To provide some idea about the potential impact of administrative fees, note that Bengen (2006a) finds a SAFEMAX of 4.15 percent for a portfolio with 50 percent large company stocks and 50 percent intermediate-term government bonds. If we include average annual administrative fees of 1.6 percent for stock mutual funds and 1.2 percent for bond mutual funds […], which we deduct at the end of each year before rebalancing, the SAFEMAX for this portfolio is reduced to 3.49 percent. Looking to the future, index funds and ETFs do provide very low administrative fees, making it more reasonable to ignore them, but retirees who invest in costly mutual funds must realize the strong impact it will have on their sustainable withdrawal rate.

TheNightsToCome
Posts: 125
Joined: Fri Jun 30, 2017 11:48 pm

Re: SWR with a real return of 1.6%

Post by TheNightsToCome » Mon Oct 30, 2017 6:55 pm

HomerJ wrote:
Mon Oct 30, 2017 3:46 pm
TheNightsToCome wrote:
Sun Oct 29, 2017 1:11 pm
siamond wrote:
Sun Oct 29, 2017 12:36 pm
TheNightsToCome wrote:
Sun Oct 29, 2017 11:03 am
I'm not sure things have ever been this "dire," given the current combination of extremely high valuations for both stocks and bonds. (LT TIPS provided a 4-4.5% real yield in 2000 when stock valuations were at record highs.) Even if so, that "n" is very small. [...]
When we include starting valuations that are below average and average, we inflate the success rate of a given withdrawal rate.
Yes, those are fair points. But the OP wasn't asking about success rates. And his math just doesn't add up, for the reasons I identified, which was the primary point I was trying to convey. The SWR doesn't change just because you may be at a high valuation point.

This being said, I would tend to agree that future returns might become more muted than they were in a fairly rosy past (at least in the US), and that, in addition, the current combination is indeed concerning. This has been discussed ad nauseam on this forum, and some folks reach the conclusion that the 'new SWR' might be 3.5% (which isn't crazy considering that other countries did experience such painful numbers in their past). But certainly not 1.6%!!
If I read Table 1 correctly (see link below to Wade Pfau study), then the SWR between 1900-2012 was less than 1.6% for seven out of 20 developed nations (and no greater than 3.68% in any nation):

https://advisorperspectives.com/art ... awal-rates

Valuations have probably never been less favorable than today, so it isn't difficult for me to imagine that the SWR will turn out to be 1.6% over the next 30 years.

Of course, those results were with 50:50 stock:bond portfolios subject to sequence-of-returns-risk. If a retiree can set up a liability matching portfolio with a guaranteed zero percent after-tax real return, then he/she can spend 3.3% of a portfolio for 30 years before running out of money. That is a higher SWR than 15/20 developed nations between 1900-2012, and only 0.37% below the SWR in the US during that period.
Wade Pfau also assumes 1% management fee, which completely skews his results compared to SatuMedia. Plus, he's been beating the doom drum since 2010, and we've gotten 12% real instead of the predicted 4% real. Yes, it's possible his models were correct, and we got really lucky... OR, maybe just maybe, its reasonable to wonder if maybe his models are missing a variable or two.

How long can PhDs make incorrect predictions before people stop listening to them? The truth is, economics really isn't a science. Too many variables (known and unknown), and the environment is constantly changing, with new laws, new industries, etc.
"Wade Pfau also assumes 1% management fee"

You are alluding to a different study. (I read that one, too.)

The study linked above simply reports the returns of 50:50 portfolios (with data from Dimson, et al), no mgmt fee (or anything else) subtracted.

It doesn't provide predictions. It simply reports historical results.

Edit: Never mind. I see now that Siamond already responded.
Last edited by TheNightsToCome on Mon Oct 30, 2017 7:03 pm, edited 1 time in total.

TheNightsToCome
Posts: 125
Joined: Fri Jun 30, 2017 11:48 pm

Re: SWR with a real return of 1.6%

Post by TheNightsToCome » Mon Oct 30, 2017 6:59 pm

siamond wrote:
Mon Oct 30, 2017 2:16 pm
TheNightsToCome wrote:
Sun Oct 29, 2017 8:53 pm
AlohaJoe wrote:
Sun Oct 29, 2017 8:41 pm
From Pfau's recent book, when taking about global SWR he writes
Bengen used intermediate-term government bonds—which seem to hold a sweet spot in the trade-off between returns and volatility—to yield the highest sustainable spending. But this bond option is not available in the Global Returns Dataset. For what is available, bills work better than long-term bonds for the aforementioned reason that their lower volatility helps more than the lower return hurts.
Well, based on this it seems likely that Pfau would choose the bills data for his SWR study. Why wouldn't he? Dimson et al have data on bills. Otherwise, he wouldn't obtain the best SWR.
I swapped a couple of e-mails with Wade, and he clarified the following:

Note that this [DMS] dataset has long-term bonds and short-term bills. Short-term bills support higher withdrawal rates and so I usually use stocks/bills now. But this is an older article where I hadn’t yet settled on that convention. Withdrawal rates are a bit less because of the long-term bonds instead of bills.

Bengen's intermediate-term government bonds is the sweet spot for the highest withdrawal rates. For the Global returns data, it is long-term bonds or bills, and bills are the better choice of the two. The long-term bonds are too volatile and too exposed to inflation.


And here is a link to a clean copy of the corresponding article about International diversification and SWRs, with tables properly formatted. Again, this was done with LT bonds at the time of writing (2014).
"Withdrawal rates are a bit less because of the long-term bonds instead of bills."

Thanks for the link. It would be interesting to know how much less.

User avatar
siamond
Posts: 3426
Joined: Mon May 28, 2012 5:50 am

Re: SWR with a real return of 1.6%

Post by siamond » Mon Oct 30, 2017 11:27 pm

TheNightsToCome wrote:
Mon Oct 30, 2017 6:59 pm
"Withdrawal rates are a bit less because of the long-term bonds instead of bills."

Thanks for the link. It would be interesting to know how much less.
I was pondering about the same question, so I gave it a try with the US returns we have in the Simba spreadsheet. Here is the outcome for various Asset Allocations (US only, stocks/treasuries). Note that an Expense Ratio is always used in such spreadsheet, e.g. modern index funds fees. Backtesting was performed starting from 1871, with the usual periods of 30 years. P10 means 10% percentile, as an attempt to not overly focus on the single worst singularity (which is usually end of the 60s). Conditional coloring is per column of 5 numbers (green being good, of course).

Image

I added the little known (and yet very meaningful imho) Ulcer Index, which is essentially a cumulative measure of depth and duration of ALL drawdowns. Here a high value is bad, so I added the 90% percentile math.

Overall, we can see that for SWRs, an LTT approach loses a quarter to half a point, while T-Bills aren't that different to ST or IT Treasuries or TBM. And when looking at both SWR and Ulcer Index as measures of long-term risk, the sweet spot could be viewed as ST Treasuries. I have no way of checking such findings with other countries though.

EDIT: updated table with a more meaningful set of AAs. Also added TBM.
Last edited by siamond on Tue Oct 31, 2017 11:29 am, edited 1 time in total.

dcabler
Posts: 290
Joined: Wed Feb 19, 2014 11:30 am

Re: SWR with a real return of 1.6%

Post by dcabler » Tue Oct 31, 2017 6:08 am

siamond wrote:
Mon Oct 30, 2017 11:27 pm
TheNightsToCome wrote:
Mon Oct 30, 2017 6:59 pm
"Withdrawal rates are a bit less because of the long-term bonds instead of bills."

Thanks for the link. It would be interesting to know how much less.
I was pondering about the same question, so I gave it a try with the US returns we have in the Simba spreadsheet. Here is the outcome for various Asset Allocations (US only, stocks/treasuries). Note that an Expense Ratio is always used in such spreadsheet, e.g. modern index funds fees. Backtesting was performed starting from 1871, with the usual periods of 30 years. P10 means 10% percentile, as an attempt to not overly focus on the single worst singularity (which is usually end of the 60s). Conditional coloring is per column of 4 numbers (green being good, of course).

Image

I added the little known (and yet very meaningful imho) Ulcer Index, which is essentially a cumulative measure of depth and duration of ALL drawdowns. Here a high value is bad, so I added the 90% percentile math.

Overall, we can see that for SWRs, an LTT approach loses a quarter to half a point, while T-Bills are quite similar to ST or IT Treasuries. And when looking at both SWR and Ulcer Index as measures of long-term risk, the sweet spot was apparently ST Treasuries. I have no way of checking such findings with other countries though.
Now Siamond, you know that good analysis always results in requests for even more analysis. :D

I wonder if what you show still holds true for something closer to a Larry Portfolio which is SCV centric? Of course it means you have to start with something like 30/70 on the left....

User avatar
siamond
Posts: 3426
Joined: Mon May 28, 2012 5:50 am

Re: SWR with a real return of 1.6%

Post by siamond » Tue Oct 31, 2017 11:47 am

dcabler's post made me realize that my selection of AAs was probably not terribly representative of what a typical retiree might do. So I updated the simulation (see this post) to use AAs ranging from 30/70 to 70/30. And added TBM while I was at it. Same conclusions.

As to dcabler's exact question (SCV/Bonds portfolios), I gave it a quick test, and I was a bit surprised to see that the previous results hold, ITT and STT appear to be the sweet spot, at least with the risk metrics I used. Now that is not the common wisdom with this type of portfolio. So I'll investigate a bit more, as a separate topic, as this departs a bit from the OP.

Post Reply