GARCH

Research on Safe Withdrawal Rates

Moderator: hocus2004

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

Post by hocus2004 »

I get a lot of e-mail asking: "How do YOU invest?"
I answer: "You don't want to know. I'm the world's worst investor!")


I especially relate to this comment. I get this one all the time too, Gummy. I hate it when people ask how I invest, and for just the reason you state above. I am NOT an investing expert, so people should not be looking at what I do as a means of determining how they should invest for themselves.

My contribution is that I am the one who discovered that it is not possible to determine SWRs without taking into account the valuation levels that apply at the start date of the retirement. That's pretty much the extent of my expertise. The fact that I figured that one out does not mean that I am the world's smartest investor. There are scores and scores of people who post at our various boards who know a lot more about all sorts of investing topics than I do. I would rank myself as about average in the depth of my overall understanding of investing topics.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

I do agree with pointing out errors in other people's work ... as well as my own!

One of the pluses I see to all of the nastiness that we have been put through is that I feel that my ideas on SWRs have been put through the toughtest test that they could possibly be put through. I thought that I was right about this stuff back in May 2002, but I was not really confident enough to go forward with it in writings that would be available to the general public. Now I possess that level of confidence. There are scores of people who would love to find holes in my approach, and the fact that they haven't been able to do so in 32 months lead me to believe that it is extremely unlikely that there are any signiicant holes to be found.

People who point out weaknesses in my arguments on a discussion board are doing me a favor. They are sparing me from the far greater embarassment of getting it wrong in a Research Report or in public speech or in a book. I will still make mistakes in my books and speeches and reports, of course. But I hope that as a result of testing my ideas on discussion boards, I will not make as many of them. The community is essentially providing me a logic check on my ideas free of charge! That's just one of the reasons why I think that this new communications medium possesses such a great power to do good in days to come.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

The so-called experts that predict doom-and-gloom or a 40,000 DOW or whatever ... they do so with complete confidence (and appalling arrogance).

I'm with you on this one too, Gummy.

This is why I am uneasy with use of the word "prediction" to describe the result of SWR analysis. All that an SWR analysis tells you is how stocks will do in the event that the future is like the past. It is of course possible that the future will not be like the past. In that event, we may see results very different from the results that SWR analysis would lead us to expect.

JWR1945 makes this point by noting that throughout history investors have been done in by surprises. We are never surprised by the same thing twice, of course. A surprise by definition is something not anticipated.

Bob Dylan sums it up well with this line from his "Love and Theft" album:

"You always need to be prepared
But you never know for what"
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

In the first part of my "sensible withdrawals" tutorial I describe the traditional mechanism for establishing a Safe Withdrawal Rate ... then say:
I submit that this is cyber-fiction and real people don't do that.


You are right about this, and not just because of the valuation matter. One of the key assumptions of the conventional methodology studies is that investors will never sell any stock regardless of whether they experience large price drops or not. This is a nonsense assumption. There is precisely zero support in the historical record for this assumption.

My sense is that you are expressing a general skepticism re SWR analysis. Wanderer is someone who very much shares this view. He of course understands that the REHP study is nonsense, but I believe that he doesn't view the Data-Based Methodology as being all that much better. I think that the Data-based Methodology is a whole bunch better, but I don't have any problem with people expressing deep skepticism re the entire SWR enterprise.

SWR analysis is a safety check. it shouldn't be your only safety check. You should always be using common sense, and you should always supplement SWR analysis with stuff you learn from lots of other sources and lots of other tools.

All that said, I believe that an analytically valid SWR study adds value. It does not solve all of your investing problems, but it provides powerful insights that can steer you in the right direction with your other investigations. There is of course nothing wrong with others putting forward a different point of view on this question.

I have a line not too far off from your cyber-fiction line. I say that the REHP study is not science, it is science fiction. I believe that the data-based studies are science, but I believe that they are scientific investigations of a limited range of questions. There is no SWR analysis that can provide you with all of the answers.
User avatar
gummy
*** Veteran
Posts: 107
Joined: Wed May 14, 2003 4:00 am
Location: Hogsville, ON
Contact:

Post by gummy »

Thanks for the history.
Do y'all have a response to what I think is my greatest concern:
... I assume that your suggestion is to ...
Change your withdrawal rate EVERY YEAR depending upon the current valuation level, namely E10/P.
Am I right in this interpretation?
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

If that's what this all comes down to, then I assume that your suggestion is to (and here's the crux of the argument, I think):
Change your withdrawal rate EVERY YEAR depending upon the current valuation level,


The SWR does indeed change with every change in valuation levels. It is constantly in flux.

But the way in which we calculate the number is to deternmine what take-out percentage is 95 percent safe for a 30-year returns sequence beginning with the valuation level that applied at a specified starting point.

For example, the SWR in January 2000 was 1.6. That is the worst valuation level we have seen. Valuations have since gone down, so the SWR has since gone up (to about 2.5 percent). But the SWR for the year 2000 retiree remains 1.6.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

Our disagreement may just be what one takes as the "valuation level", eh what?

There is of course lots of room for reasonable differences of opinion on this aspect of the question. Bernstein does it one way, raddr another, and JWR1945 another. All three approachs are analytically valid, in my view. And I have no doubt but that there are other options that are also analytically valid.

My personal preference is the JWR1945 methodology. But I say "Let a Thousand Flowers Bloom" on this one.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

I think it's inane to do studies that claim to PROVE (?) that a certain scheme is best, based upon regression lines, standard deviations, correlations and all that mathemagical paraphernalia as it pertains to historical data.

I'd prefer some logical arguments, based upon common sense. THEN see if this common sense strategy would have worked in the past.


We are in agreement on this one. Common sense trumps statistics every time in my book.

The big plus of doing a statistical anlaysis is that it may point out to you flaws in thinking that you believe to be the result of common sense. There are many people today who believe that it is "common sense" that long-term timing is not possible. They are wrong, for reasons that Bernstein explains in depth in his book "The Four Pillars of Investing."

You shouldn't allow what you learn from statistical research to trump your common sense. That is madness. But you should aim to remain open to new ideas, and new ideas can be uncovered through the use of statistical analysis which generates findings not in line with what you once thought was a common-sense belief.
User avatar
gummy
*** Veteran
Posts: 107
Joined: Wed May 14, 2003 4:00 am
Location: Hogsville, ON
Contact:

Post by gummy »

realize that "The SWR does indeed change with every change in valuation levels. It is constantly in flux"
... but what I was interested in is the question:

Does Sally (our typical investor) change her withdrawal every year (every month?) depending upon the current E10/P level?

If so, is she to look at the historical relationship between E10/P vs SWR and pick her current withdrawal rate from that relationship?

And what if she doesn't know the E10/P for her 100% GE portfolio?

Although the high correlation between E10/P and a 80% / 20% portfolio is fascinating, I don't think it's easy to devise a strategy which takes this into account.
User avatar
gummy
*** Veteran
Posts: 107
Joined: Wed May 14, 2003 4:00 am
Location: Hogsville, ON
Contact:

Post by gummy »

hocus:
Here's an interesting scenario:
First, look at E10/P for our 80% / 20% portfolio (1928 - 2000)


The average is 7.0%.

Now, assuming that we have one of those 80% / 20% portfolios, we want to increase or decrease our annual withdrawals based upon the current E10/P value.

To this end, we do this:
1: Look at the "standard, garden variety 4% withdrawal RATE" (changing with inflation).
2: Add E10/P -7.0% to this RATE (so it changes annually according to the current E10/P value, increasing when the E10/P valuation is larger than average and decreasing when it's lower).
3: Finally, we take 90% of this modified rate, giving a recommended Withdrawal Rate for the current year (for Sally).

Since 1966-1995 was perhaps the worst 30-year period for that "standard, garden variety 4% withdrawal RATE", let's compare our new withdrawal rate proposal to the (disastrous) 4% thing:


Note that, whereas the "4% SWR" withdrawals continue to increase with inflation, our E10/P withdrawals increase or decrease according to the current E10/P ... and (as we'll see below) our portfolio survives :)

Here's a picture of the two portfolios:
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

Does Sally (our typical investor) change her withdrawal every year (every month?) depending upon the current E10/P level?

Once she is retired, she does not need to make any changes in the withdrawal rate if she uses JWR1945's numbers because JWR1945's methodology determines the number that works in a worst-case scenario (assuming that the future is no worse than the past).

If she is not yet retired, then she needs to determine what the SWR is on the date of her retirement in order to have a number that will work in a worst-case scenario.

You can't say in advance what the SWR is going to be on your retirement date because the SWR changes with changes in valuation levels. You can make an assumption as to what the valuation level will be, and look up what SWR would apply at that valuation level. But until you know for sure what valuation level applies at the start of your retirement, you don't know what SWR applies for your particular retirement.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

Here's an interesting scenario:

I appreciate you doing those charts, Gummy. I'm not capable of understanding what is going on with them. It is one of the great ironies of this entire matter that a guy with no numbers skills whatsoever happened to play such a big role in it. I really can only understand things when they are translated into words. I always cheat with the JWR1945 posts. I skip all the numbers stuff and try to figure out what is going on by reading the part where he summarizes the results in words.

Perhaps JWR1945 will have a chance to look at these charts in a bit and he will be able to offer some useful feedback on them
User avatar
gummy
*** Veteran
Posts: 107
Joined: Wed May 14, 2003 4:00 am
Location: Hogsville, ON
Contact:

Post by gummy »

Once she is retired, she does not need to make any changes in the withdrawal rate
Aah, I see.
I was beginning to think that this E10/P stuff was VERY similar to the "sensible withdrawals" strategy (which suggests that you change your withdrawal rate each year).

In the above charts, had you withdrawn at that "standard 4%" rate (that everybuddy talks about), then your $100K portfolio would drop dead in 27 years.

Had you withdrawn at a rate determined by the current E10/P valuation, you'd have over $100K by year 30.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

I was beginning to think that this E10/P stuff was VERY similar to the "sensible withdrawals" strategy (which suggests that you change your withdrawal rate each year).

I'm beginning to think that our two approaches were just different ways of addressing a similar core concern. We both sensed that the claims made by conventional methodology study enthusiasts of "100 percent safety" were not right. You responded with an approach that calls for changes throughout the retirement in the take-out number used, and I responded with an approach that calls for use of a stable number but a lower number than the one identified as "100 percent safe" in the conventional methodology studies.
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

I have modified a calculator, which I have named Gummy 01, with the following version of Gummy's new algorithm:
Withdrawal Amount = Current Balance*(Multiplier)*(100E10/P+Investment Expenses input)

I set the Multiplier to 0.9.
I set the Investment Expenses input (in cell B15) equal to -3.00%.
I selected 80% stocks and 20% 30-year Treasury Bonds.
[I rebalanced annually. Withdrawals were split equally between the beginning and the end of each year. The CPI was used for making inflation adjustments.]
I set the standard withdrawal rate to zero percent of the initial balance.

Here are my 1966 (real) balances from 1966-2002:

Code: Select all

100000
91307
98546
101757
88751
87453
92390
100436
78672
55973
65193
64849
53859
52379
48992
49825
40780
45263
48150
47157
52551
63040
58557
62566
69330
64436
77774
79408
84351
82454
102495
121456
147064
183158
203807
192368
Balance at year 30, which is 1996: 102495.

Here are the amounts withdrawn in real dollars.

Code: Select all

1665
1438
1575
2466
2476
2318
2080
3517
4911
3303
3436
4243
3762
3743
3466
4354
3385
3083
3104
2559
1997
2349
2019
1730
2079
1353
1389
1263
1497
886
545
54
(816)
(1271)
(537)
508
You had to put more money into the account in 1999-2001!

I will start on Gummy 02 tomorrow. It will vary the amount withdrawn according to the following algorithm:
Withdrawal Amount = Initial Balance*(Inflation Adjustment)*(Multiplier)*(100E10/P+the standard Withdrawal Rate input)

We will be able to combine the two algorithms.

My data show a balance that drops before 1996, which is year 30, and then starts to increase above the initial balance. Gummy's graph shows the opposite. He may be using the Gummy 02 algorithm while I am using the Gummy 01 variant.

Have fun.

John R.
User avatar
gummy
*** Veteran
Posts: 107
Joined: Wed May 14, 2003 4:00 am
Location: Hogsville, ON
Contact:

Post by gummy »

John
When I clean up my spreadsheet** I'll stick it on my website.

What I did was "pretend" I was doing the 4% so-called SWR and added (each year) last year's E10/P - 7.0%, then scaled down to 90%.

Example for 1974:
The 4% SWR was 6.5% (of the initial portfolio).
That's 4%, increased by 9 years of inflation.

E10/P was 5.9%

The "calculated withdrawal" is then:
0.9 (6.5 + 5.9 - 7.0) or 4.8%

(which is significantly less than the 4%SWR of 6.5% as shown on the 2nd chart above).

** The spreadsheet is in such a mess (and no doubt has bugs), but the idea will remain the same when it's neat and tidy.
http://home.golden.net/~pjponzo/E10-P-SS.gif

NOW, here's burning question #1:
What does Sally do if her portfolio is 100% GE stock?

Poor Sally hasn't the faintest idea of historical E10/P for GE.

... and burning question #2:
If Sally's $100K portfolio becomes $500K after ten years, does she still withdraw at the constant rate determined when she started?

... or can she take that trip to Hawaii?
User avatar
gummy
*** Veteran
Posts: 107
Joined: Wed May 14, 2003 4:00 am
Location: Hogsville, ON
Contact:

Post by gummy »

P.S.
I'm fiddling with (yet another!) tutorial, modifying the standard, garden variety SWR (like the notorious 4%) to take into account something like the current E10/P (compared to some average E10/P level).
http://home.golden.net/~pjponzo/E10-P.htm

The problem?
It'll be of little value unless the user has access to E10/P data for her particular portfolio
Mike
*** Veteran
Posts: 278
Joined: Sun Jul 06, 2003 4:00 am

Post by Mike »

It looks like you could badly lag inflation with this system. As a ball park rule of thumb, it looks like you have to be prepared to cut out 1/3 or so of your original expenses.
Mike
*** Veteran
Posts: 278
Joined: Sun Jul 06, 2003 4:00 am

Post by Mike »

It'll be of little value unless the user has access to E10/P data for her particular portfolio
One additional factor to consider. Even when a person can calculate their E10/P, as a general rule portfolios less diversified than an S&P index can safely withdraw less for the same E10/P.

http://www.retireearlyhomepage.com/concport.html
User avatar
gummy
*** Veteran
Posts: 107
Joined: Wed May 14, 2003 4:00 am
Location: Hogsville, ON
Contact:

Post by gummy »

... as a general rule portfolios less diversified than an S&P index can safely withdraw less for the same E10/P
Amen.
It'd be nice if one could INSIST that all investors diversify, however some companies (like that of my son-in-law) will match employee contributions provided they invest in company stock :?

P.S.
The spreadsheet (such as it is), is here:
http://www.gummy-stuff.org/SS/E10-P.xls
Post Reply