Getting a newbie up to speed quickly

Research on Safe Withdrawal Rates

Moderator: hocus2004

Norbert Schlenker
* Rookie
Posts: 43
Joined: Mon Mar 21, 2005 4:00 am
Location: The Dry Side of the Wet Coast

Getting a newbie up to speed quickly

Post by Norbert Schlenker »

I've read the announcements and the stickies at the top of the forum and I'm going to ask some clueless newbie questions. I refer first to this thread, which JWR1945 says "settles the great debate".

1. What's the great debate?
2. Does it have anything to do with the dreaded 4%?
3. If so, using JWR1945's tables in the linked thread, I see no failures at anything under 4.2%, not even when P/E10 is as high as 27. If that's so, shouldn't 4% be safe as long as P/E10 is 27 or less?
Great minds think alike. Fools seldom differ.
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

Norbert Schlenker wrote:I've read the announcements and the stickies at the top of the forum and I'm going to ask some clueless newbie questions. I refer first to this thread, which JWR1945 says "settles the great debate".

1. What's the great debate?
2. Does it have anything to do with the dreaded 4%?
3. If so, using JWR1945's tables in the linked thread, I see no failures at anything under 4.2%, not even when P/E10 is as high as 27. If that's so, shouldn't 4% be safe as long as P/E10 is 27 or less?
The great debate was whether Safe Withdrawal Rates depend upon stock valuations. They do. That is where the debate ends.

[The great debate was usually sidelined by tactics to hinder any kind of research along these lines. Those tactics were the Debate about having a Debate.]

The claim had been made and it continues to be made that a 4% withdrawal rate was always safe (that's right: 100% safe) when using a portfolio of 74% (or something close) stocks as represented by the S&P500 index and commercial paper. Allowances are made for unlikely exceptions such as having an asteroid collide with the earth (within the retirement period).

No. The Historical Surviving Withdrawal Rates (previously known as Historical Database Rates) tell us only what has happened. They do not tell us what is likely to happen. To make a projection, it is necessary to introduce statistics and confidence limits.

Even crude confidence limits are vastly superior to having no confidence limits at all.

There is solid evidence that the historical sequences in the past were lucky sequences or, at least, far from being unlucky. We have been able to quantify such effects.

Portfolio survivability is influenced by both the overall stock market returns and the actual sequence of returns. This widens out the range of withdrawal rates that are likely to survive. No, the actual Historical Surviving Withdrawal Rates did not cover the range of uncertainty. What is worse, only a very few data points are directly relevant if the relationship with valuations is not taken into account. Instead of having fifty to a hundred relevant sequences, you only have four or five, which is far too low a number.

But what is even more important is that we are still in bubble territory. Stocks from those historical sequences had dividend yields above 2.9% and usually much higher. Valuations in the historical record peaked at P/E10 = 27. At the peak of the bubble, P/E10 = 44. Today, P/E10 = 28 to 29.

None of the results in the historical record apply directly to today's stock market.

We can extrapolate into today's valuations reasonably well using the relationship between Safe Withdrawal Rates (which are the lower confidence limit about the Calculated Rate produce by Excel curve fitting) and the percentage earnings yield 100 / [P/E10]. P/E10 is the current price of the S&P500 index and E10 is the average of the trailing ten years of earnings.

Have fun.

John R.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

I agree with what was said in the JWR1945 post above. I'll add a few comments of my own here.

What's the great debate?

Intercst founded the first Retire Early board (the Motley Fool board). Because he was the author of an SWR study (the study published at RetireEarlyHomePage.com), using SWR analysis to plan an early retirement became the accepted thing to do in our community. Thus, when I came forward in May 2002 with a series of posts showing that intercst got the number wrong in the study, it upset our apple cart quite a bit.

We would NOT have seen the level of negativity we have seen had intercst reacted in a responsible manner. There were many community members who expressed great excitement about learning about the Data-Based SWR Tool when I first put the idea forward. The trouble started when intercst began engaging in deception and intimidation to block reasoned debate. That caused scores of our best community members to leave the community, and allowed the voice of the intercst wing of the community to become dominant in the discussions.

I call the discussions "The Great SWR Debate" because of the reaction we have seen to my May 13, 2002, post. The SWR matter has been the dominant topic for discussion at six boards for close to three years now. It has generated tens of thousands of posts. Obviously there is more than a little interest in our community in knowing what the historical data really says re SWRs.

Does it have anything to do with the dreaded 4%?

The Data-Based SWR Tool has discredited conventional methodology studies, which report a 4 percent take-out to be safe regardless of the valuation level that applies at the start-date of the retirement.

If that's so, shouldn't 4% be safe as long as P/E10 is 27 or less?

The answer is "no." My sense is that this is a point that many people still do not get.

The flaw in the conventional methodlogy was its failure to take changes in valuation into account. This analytical flaw became a far more serious problem when valuations reached the levels they reached in the late 1990s. But the flaw was always present. The conventional methodology always reports the wrong number as the SWR, except in that once-in-a-blue-moon case where the correct number just by coincidence happens to be 4.

If someone asked you, "What is 4 minus 2," and you responded "4," would you be correct? If they said that you were wrong, would it be an effective response for you to say "4 really is one of the numbers you put forward, prove to me that you never mentioned the numeral 4!" would that response be a reasonable one? I say no.

The same is so if someone asks you "What is 4 minus 1?" or "What is 4 plus 1? or "What is 4 plus 2?" It is always necessary when answering a question under consideration to look at all the factors with a critical bearing on the answer to that question.

If someone asks you "What is the historical surviving withdrawal rate" you would be correct to say "4". That is the number you get for the HSWR when you look at the historical data. But the HSWR is NOT the same thing as the SWR. There are times when it is safe to use the HSWR as your take-out number and there are times when it is risky to use the HSWR as your take-out number. There are also times when the historical data indicates that it is safe to use a take-out number well in excess of 4 percent.

If you want to know what take-out number is safe according to the historical data, you need to look at the factors that the historical data says affect safety. The single most critical factor is the valuation level that applies on the date the retirement begins. You can't possibly determine what take-out number is safe without taking that factor into account. The REHP study does not take that factor into account. The methodology used in that study is analytically invalid for purposes of determining SWRs
beachbumz
** Regular
Posts: 79
Joined: Mon Feb 07, 2005 4:00 am
Location: Cocoa Beach, FL

Post by beachbumz »

JWR1945 wrote:The claim had been made and it continues to be made that a 4% withdrawal rate was always safe (that's right: 100% safe) when using a portfolio of 74% (or something close) stocks as represented by the S&P500 index and commercial paper. Allowances are made for unlikely exceptions such as having an asteroid collide with the earth (within the retirement period).

Have fun.

John R.
Hi JWR!

I'm a little confused here. Are you saying that the 4% rate was NOT always safe when back tested in the PAST. IOW, was there any point in the past (that was covered by the REHP study) that if I had retired and withdrew 4% that I would have run out of money (in a 30 yr period)? I assume the lastest date I could have 'retired' would be 1970ish since the study didn't go past 2000 (I believe).

Thanks,

Beachbumz 8)
Life is Good.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

Are you saying that the 4% rate was NOT always safe when back tested in the PAST.

The 4 percent number was not always safe in earlier periods, BeachBumz. It always SURVIVED in earlier periods. That's something different.

Say that I have driven 10 miles to a friend's house every Sunday night for two years now. That's 104 drives, 208 if you count back and forth.

On two of those occasions, I was a little bit drunk on the return trip. I got in a fender bender on one of those drives and was hospitalized after a car crash suffered during the other. In both cases, I lived.

Is it reasonable to say that I was "100 percent" safe on the two trips I took when drunk just because I happened to survive them? It is not.

I was taking a big chance driving drunk. I got lucky. I survived. That doesn't mean that it is "100 percent safe" for me to drive home drunk today.

The portfolios of investors who retired with 74 percent S&P allocations in the late 20s or the mid-60s survived, according to the historical data. These retirements were not "100 percent safe," according to the historical data.

The term "100 percent safe" has a defined meaning in SWR analysis. It means that the portfolio works in a worst-case returns sequence (the worst seen in the historical record, but not any worse than that). Had worst-case scenarios popped up for those retirees, their plans would have failed. It turned out that they didn't fail because worst-case return sequences did not in fact turn up.

Those retirees were lucky. That's nice for them. But it is not reasonable to conclude that because they happened to get lucky that is "100 percent safe" to follow in their footsteps.

There are some valuation levels at which a 4 percent take-out is safe according to the historical data and there are some at which it is not. To find out whether that take-out number is safe or not, you need to look at the historical data using a methodlogy that is analytically valid for purposes of determining SWRs.
User avatar
ben
*** Veteran
Posts: 231
Joined: Mon Feb 17, 2003 4:00 am
Location: The world is not enough

Post by ben »

BB;
I believe both Intercst and Dory have updated nos up to and including 2003 or 2004. So you can get an exiting historical insight to the first years of the new millinium too.

But historically none of the portfolio went bust at 4%. Now for the future...
who knows? Anyway only fools would not diversify more, adjust w/r depending on market Etc.

Cheers!
Normal; to put on clothes bought for work, go to work in car bought to get to work needed to pay for the clothes, the car and the home left empty all day in order to afford to live in it...
User avatar
ben
*** Veteran
Posts: 231
Joined: Mon Feb 17, 2003 4:00 am
Location: The world is not enough

Post by ben »

The 4 percent number was not always safe in earlier periods, BeachBumz. It always SURVIVED in earlier periods. That's something different.
I think this is a key point in some misunderstandings of your posts hocus. I read safe/survived as the same thing - you do not. We don't have to agree.
Cheers!
Normal; to put on clothes bought for work, go to work in car bought to get to work needed to pay for the clothes, the car and the home left empty all day in order to afford to live in it...
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

But historically none of the portfolio went bust at 4%

You're not shooting straight, Ben. Your aim is crooked..
unclemick
*** Veteran
Posts: 231
Joined: Sat Jun 12, 2004 4:00 am
Location: LA till Katrina, now MO

Post by unclemick »

And then there are old school blockheads like me who watch dividends and interest - interest and dividends - aka the income stream.

Extreme diversification with a value twist is a viable alternate route - provided you keep expenses low, don't overtrade - er rebalance I mean -

AND WATCH THE INCOME STREAM

Heh, heh, heh, heh, heh!

Blockheads of the world unite! Three cup of coffee post.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

I read safe/survived as the same thing - you do not. We don't have to agree.

You need to get yourself up to speed on the ABCs, Ben. They are not even close to being the same thing.

The difference between using a withdrawal rate that the historical data reveals survived historically and using a withdrawal rate that the historical data reveals as safe at today's valuation levels is the difference between a retirement plan that works and a retirement plan that goes bust. That's a significant difference.

Please put some effort into learning the ABCs before putting up your next post, Ben. Or else identify yourself in your next post as a newbie to SWR analysis asking an extremely basic question. You put retirements at risk when you employ a tone suggesting that you possess some knowledge of the subject matter when in fact you do not.
User avatar
ben
*** Veteran
Posts: 231
Joined: Mon Feb 17, 2003 4:00 am
Location: The world is not enough

Post by ben »

You need to get yourself up to speed on the ABCs, Ben

He,he - yes might be caused by me not being a native English speaker! :D
Normal; to put on clothes bought for work, go to work in car bought to get to work needed to pay for the clothes, the car and the home left empty all day in order to afford to live in it...
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

might be caused by me not being a native English speaker!

You speak intercst, do you? I understand that Dory36 has been taking some classes..
unclemick
*** Veteran
Posts: 231
Joined: Sat Jun 12, 2004 4:00 am
Location: LA till Katrina, now MO

Post by unclemick »

My two cents.

Of course Ben and Intercst and Raddr are shooting straight.

The central difficulty arises from the appearance that circa 2000 the historical trend channel was broken to the upside(in the U.S.). Add in the current low interest rate and things get chewy.

The insights provided by SWR reasearch data runs along with the nuts and bolts of calculator limitations - provide opportunities to examine how to play the game today.

More than one way to skin a cat - real estate, global diversification, dividend strategies, TIPs anchors, the quest for classic value, etc., etc.

The debate is endless - at least for as long as I can remember(I started in 1966) and will probably go on after I meet my reward.

The SWR research to date has sharpened my sense of how I want to proceed for my ER and clarified some of my thinking.

Number on!
Last edited by unclemick on Tue Mar 22, 2005 7:30 am, edited 2 times in total.
Norbert Schlenker
* Rookie
Posts: 43
Joined: Mon Mar 21, 2005 4:00 am
Location: The Dry Side of the Wet Coast

Post by Norbert Schlenker »

JWR1945 wrote:The great debate was whether Safe Withdrawal Rates depend upon stock valuations. They do. That is where the debate ends.
Let me be just a little blunt. That flat assertion has the tone of religious fervor. I assume this debate was conducted somewhere, probably online, and that you could provide a pointer to it.
[The great debate was usually sidelined by tactics to hinder any kind of research along these lines. Those tactics were the Debate about having a Debate.]
I will keep that in mind and discount the meta-debate.
The Historical Surviving Withdrawal Rates (previously known as Historical Database Rates) tell us only what has happened. They do not tell us what is likely to happen. To make a projection, it is necessary to introduce statistics and confidence limits.

Even crude confidence limits are vastly superior to having no confidence limits at all.
Of course. However, you shouldn't just ignore the fact that the difference between the 4% HSWR and the actual historical returns is also a "crude confidence limit". I don't want to duplicate my post in the other thread about analytical techniques, so won't comment further here.
There is solid evidence that the historical sequences in the past were lucky sequences or, at least, far from being unlucky. We have been able to quantify such effects.
I'm a little surprised that you can quantify good luck. Is there a thread where this quantification is demonstrated?
Portfolio survivability is influenced by both the overall stock market returns and the actual sequence of returns. This widens out the range of withdrawal rates that are likely to survive. No, the actual Historical Surviving Withdrawal Rates did not cover the range of uncertainty.
I agree with the first two sentences. What justifies the leap to the third?
What is worse, only a very few data points are directly relevant if the relationship with valuations is not taken into account. Instead of having fifty to a hundred relevant sequences, you only have four or five, which is far too low a number.
Your small sample criticism is justified. Can you demonstrate that tying a safe withdrawal rate to initial valuation is similarly justified?
Valuations in the historical record peaked at P/E10 = 27. At the peak of the bubble, P/E10 = 44. Today, P/E10 = 28 to 29. None of the results in the historical record apply directly to today's stock market.
The difference between 27 and 28-29 is likely in the noise. So the 1929 experience with a P/E10 of 27 is likely to be a good proxy for today's situation. I understand that it's a problem for your thesis that a 4% SWR beginning in 1929 survived. You need to address it, not brush it off.
We can extrapolate into today's valuations reasonably well using the relationship between Safe Withdrawal Rates (which are the lower confidence limit about the Calculated Rate produce by Excel curve fitting) and the percentage earnings yield 100 / [P/E10]. P/E10 is the current price of the S&P500 index and E10 is the average of the trailing ten years of earnings.
What makes you so sure that we can extrapolate? I really would prefer to talk pure stats in the analytical thread, but as this is here, let me ask this question. When you run your regressions to calculate coefficients and confidence intervals, what sort of r^2 and significance values are you getting from the regressions? I apologize if there is a thread among the hundreds here where this is discussed. I couldn't find one using the board's search facility so a pointer would be great.
Great minds think alike. Fools seldom differ.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

Of course Ben and Intercst and Raddr are shooting straight.

This isn't a joke, UncleMick.

Dory36 has compromised the integrity of discussions held at his site--which is a very important site--so that a con man can hang on a few months longer.

If you're still laughing, you need to stop laughing and start thinking.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

I understand that it's a problem for your thesis that a 4% SWR beginning in 1929 survived. You need to address it, not brush it off.

Most of the questions in this post are for JWR1945 and most are reasonable questions. In this one we see the attitude pop to the surface again. It is a dead giveaway of someone who wants to defend intercst, feels frustrated that the nonsense doesn't add up, and has no other tricks available to him but the low one of loading a post with attitude and misrepresentation.

There is no problem at all that a 4 percent take-out is the historical surviving withdrawal rate (HSWR). That is a fact. How could a fact be a problem for someone doing legitimate research?

Facts are problems for intercst supporters, not for those taking an honest look at the historical data to determine what it says re what withdrawal rate is safe.

How did you get the idea in your head that the fact that a 4 percent take-out survived for a retirement beginning in 1929 is a problem for an advocate of the Data-Based SWR Tool, Norbert?
unclemick
*** Veteran
Posts: 231
Joined: Sat Jun 12, 2004 4:00 am
Location: LA till Katrina, now MO

Post by unclemick »

I stand by my post.

And I think Bernstein may be full of it from time to time.

I respect his numbers - but I came to an entirely different conclusion than he did in his 15 Stock Diversification Myth.

Maybe some alpha male hormones clouding my judgement - but there it is.

Number on - keep up the good data runs.

P.S.: Unclemick's mish mash portfolio marked to market - ballpark 3% now a days. I can live with that - for my ER.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

Can you demonstrate that tying a safe withdrawal rate to initial valuation is similarly justified?

Again, it is for JWR1945 to respond to the statistics questions. But I believe that the table below (which he prepared) might help with the above.

The table compares the PE10 at the start of a 30-year historical sequence and the highest surviving withdrawal rate that ultimately applies for that sequence (we refer to this number as the historical database rate, or HDBR--that's the same thing as the historical surviving withdrawal rate, or HSWR).

HDBR50 and HDBR80 Ordered by PE10

Lowest Valuations

Code: Select all

 
Year  PE10  HDBR50  HDBR80  
1921    5.1   9.6   11.6  
1922    6.2   8.4   10.3  
1924    8.0   7.9    9.5  
1923    8.1   7.6    9.0  
1933    8.7   5.8    8.4  
1980    8.8   9.2   10.3  
1975    8.9   7.0    8.3  
1978    9.2   7.9    9.1  
1979    9.2   8.4    9.5  
1932    9.3   6.0    8.1  
1925    9.6   7.3    8.4  
1942   10.1   6.1    9.1  
1943   10.1   6.3    9.2  
1949   10.2   7.8   11.1  
1948   10.4   7.7   10.9  
1950   10.7   7.3   10.2  
1944   11.0   6.1    8.6  
1976   11.1   6.5    7.2  
1926   11.3   6.9    7.7  
1935   11.4   5.3    7.4 


Middle Valuations

Code: Select all

 
Year  PE10  HDBR50  HDBR80  
1947   11.4   6.8   9.4  
1977   11.4   6.8   7.4  
1951   11.8   7.1   9.4  
1945   11.9   5.9   8.2  
1954   12.0   6.9   9.0  
1952   12.5   6.9   9.0  
1934   13.0   4.8   6.3  
1953   13.0   6.7   8.6  
1927   13.1   6.6   7.3  
1938   13.5   4.9   6.6  
1974   13.5   5.6   5.9  
1958   13.7   5.8   6.8  
1941   13.9   5.0   7.0  
1939   15.5   4.6   6.0  
1946   15.6   5.2   6.7  
1955   15.9   5.8   6.9  
1940   16.3   4.6   6.1  
1971   16.4   4.9   5.0  
1931   16.7   5.1   5.6  
1957   16.7   5.3   6.0  


Highest Valuations

Code: Select all

 
Year  PE10  HDBR50  HDBR80  
1936   17.0   4.4   5.5  
1970   17.0   4.8   4.9  
1972   17.2   4.8   4.8  
1959   17.9   5.0   5.4  
1956   18.2   5.2   5.9  
1960   18.3   5.0   5.3  
1961   18.4   5.0   5.3  
1973   18.7   4.7   4.5  
1928   18.8   5.7   5.8  
1963   19.2   4.9   5.1  
1967   20.4   4.5   4.5  
1962   21.1   4.7   4.8  
1969   21.1   4.3   4.2  
1968   21.6   4.3   4.2  
1937   21.6   3.9   4.6  
1964   21.6   4.6   4.6  
1930   22.3   4.8   4.8  
1965   23.2   4.3   4.2  
1966   24.0   4.2   4.0  
1929   27.0   4.6   4.3  

I see a strong correlation between the PE10 that applies at the start of a sequence and the HDBR that can only be determined 30 years later. It's important to understand that, in the event that there is any correlation whatsoever, the conventional methodology numbers are wrong. The conventional methodology makes no adjustment whatsoever for changes in valuation levels. It churns out the same SWR no matter how high or low current valuations happen to go.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

And I think Bernstein may be full of it from time to time.

That's a fair comment. I have no problem at all with that one.

But if some community members believe that Bernstein got the SWR right and some think he is full of it. then both groups should be able to engage in a reasoned exchange of views at the Early Retirement Forum. Dory36 has made clear with his recent actions that the penalty for believing that Bernstein got the number right and intercst got it wrong (and expressing that view in honest posts put to his forum) is the death penalty--expulsion from the site.

Dory36 has compromised the integrity of his discussion boards in a serious way. And I think it is safe to say that he did this on intercst's behalf. It is clear from a number of comments that Dory36 has put forward that he doesn't for 10 seconds share intercst's views on SWRs or on what are acceptable posting practices.

So we now have another once well-respected community member now headed down the path of deception on behalf of a con man who happened to be the person who founded the first board. Does there come a time when the special status accorded to those who founded first boards expires? Does there come a time when respected community members start developing a reasonable concern for their own reputations and for the growth of our boards and for the success of the retirement plans developed by those making use of the boards?

I think that a good time for us to begin making a shift in priorities might be--

Right About Now!
Norbert Schlenker
* Rookie
Posts: 43
Joined: Mon Mar 21, 2005 4:00 am
Location: The Dry Side of the Wet Coast

Post by Norbert Schlenker »

hocus2004 wrote:I understand that it's a problem for your thesis that a 4% SWR beginning in 1929 survived. You need to address it, not brush it off.

In this [statement] we see the attitude pop to the surface again. It is a dead giveaway of someone who wants to defend intercst, feels frustrated that the nonsense doesn't add up, and has no other tricks available to him but the low one of loading a post with attitude and misrepresentation.
What defense of intercst? What frustration? What attitude? What misrepresentation?
How did you get the idea in your head that the fact that a 4 percent take-out survived for a retirement beginning in 1929 is a problem for an advocate of the Data-Based SWR Tool, Norbert?
Your thesis, summarized to one sentence, is that a safe withdrawal rate depends on initial valuation. In other threads, JWR1945 states explicitly that the current SWR is very low, much below 4%, because the current valuation is very high. He bases that conclusion on regressions (which seem at first glance to have methodological problems but we can discuss that elsewhere) which include the 1929 starting point.

A portfolio with a 4% SWR starting in 1929 did not fail. It didn't fail because 4% is far below the portfolio's subsequent long run average return. So what's different today? It isn't me who identified the independent variable in your regressions, Shiller's P/E10. Whoever chose it, it's chosen and you have to live with the consequences of choosing that particular variable when you try to extrapolate from what the resulting regression tells you. You can't just buy the pieces you like - lower SWRs and wider confidence intervals - and ignore the pieces you don't like - the 1929 starting point survived and today's P/E10 is very similar.
Great minds think alike. Fools seldom differ.
Post Reply