A slightly later version of the Trinity study is available at

http://www.afcpe.org/doc/Vol1014.pdf

Most withdrawal studies use a roughly 60/40 market weight allocation typical of balanced funds, with the higher equity amount needed to provide the later life growth needed to offset inflation - lower equity allocations did not historically support an inflation adjusted withdrawal.

In the summary of the above study, the authors suggest lower a starting withdrawal for longer periods, higher for shorter. Simply dividing your portfolio annually over your IRS life expectancy will give a 4% withdrawal at 25 years (3.3% at 30, 5% at 20), eventually resulting in a late life spend down.  This has  the advantage over spending a fixed percent of portfolio annually (3, 4, 5, 20, 90 all work, although the last two rapidly deplete your portfolio) in that it links the withdrawal to life expectancy.

James Walter Russell 

You can continue to post without research if it pleases you, but if you had read Trinity, the original SWR study, you would understand at that at least two of your assertions are false. Trinity is based on an uderstanding of probabilities which you fail to achieve on your site. Furthermore, the "S" in SWR doesn't mean "Safe". In this world nothing, especially investing, is "safe". All investments have risk, btw, a probablistic concept.

cheer up

j  

rmark1 wrote the following post at 05-21-2008 7:25 PM:

Rmark - My comments about research were were not directed at you. They referred to references by JWR's web site - information that is neither referenced nor peer-reviewed.

OTOH your comments are well stated,and well referenced with limitations clear.

Thanks for an excellent example.

j   

JWR

615 pages just tells us you are a programmer.

As a research scientist can you tell me  the Trinity page that mentions Safe Withdrawal Rates?

keep on trucking

j

An inflation adjusted withdrawal is only available if you hold a portfolio which produces it.

-Withdrawal studies have focused on US securities, arriving at past 4% inflation adjusted withdrawals from a roughly 60/40 market weight portfolio over 30 year periods, with about a 90% success rate.

Sustainable Withdrawal Rates From Your Retirement Portfolio (1999)

Philip L. Cooley, Carl M. Hubbard and Daniel T. Walz

-However there is a whole world out there with lower returns and higher volatility. These are usually aggregated together as ‘international’, which de-emphasizes their individual nature. Lower return + more volatility suggests a lower (or more variable) withdrawal rate than US only data.

Irrational Optimism (2003)

Elroy Dimson, Paul Marsh, and Mike Staunton

International diversification and retirement withdrawals (2003?)                                                              Danny Ervin, Larry Filer, Joseph Smolira

-Another problem in using the academic studies in real life is investor actions. People tend to invest more (less) at the peak (trough) of the market resulting in returns lower than the market return assumed in most studies.

Dumb money: Mutual fund flows and the cross-section of stock returns (2006)

Andrea Frazzini, Owen A. Lamont

What are stock investors actual historical returns? (2004)

Ilia Dichev

-Historically you had two investing choices – actively managed mutual funds or a personal portfolio of relatively few securities. Current active management likely costs at least 1.5%, and there is no reason to believe past costs were lower. And a small personal portfolio may have had wildly non-average results.

The broad based low cost index fund with near market returns was simply was not available in the past.

John Bogle, see his speeches linked through Vanguard.com

In my opinion -the 4% of initial portfolio + annual inflation looks terribly weak once the data is expanded to include other individual markets, poor timing of investors, and actual past costs are introduced. I personally would not rely on it. I suspect Henry Hebeler’s assumption of 0% after tax net real retiree returns is closer to the historical facts. And the math is much simpler for an uninterested spouse.

JWR1945a:

They Got It Wrong 

The original Safe Withdrawal Rate studies broke new ground. They advanced our knowledge of retirement finances tremendously. Yet, it is amazing how many things they got wrong.

http://www.early-retirement-planning-insights.com/they-got-it-wrong.html

 Have fun.
John Walter Russell

 John,
Here is the content of one paragraph under the page you linked to, authored solely by you. I have numbered the statements for ease of reference. I have a couple of questions after the paragraph. 

**********************************************************
Title: What They Got Wrong

#1 Early researchers tried to avoid probability and statistics.

#2 This leads to serious errors as valuations approach extremes, as they do today.

#3 Early researchers did not attempt to find a measure of valuations, nor did they seek to exploit such a measure.

#4 Yet, introducing valuations is the most important factor by far.

#5 Early researchers treated bonds simply as single year trading vehicles, sometimes including the effect of capital gains and losses, but more often simply varying the single year interest rate.

#6 This led them to overlook the true benefits of fixed income securities.

#7 They missed out on powerful alternatives to stock portfolios.

#8 Early researchers rebalanced portfolios annually.

#9 This eliminates the upside and offers very little protection on the downside.

#10 Compared to using valuations, this is a horrible mistake.

#11 Early researchers estimated the 30-Year Safe Withdrawal Rate to be 3.9% or 4.0% (plus inflation).
 **********************************************************

JWR: The numbers below correspond to the numbered items, above:

 #1. Please give some sort of identity to these "Early researchers" so that we can research the accuracy of your claims. Use of such nebulous, non-traceable, non-specific  terms as 'early researchers', when used to deride things as 'horrible',  'serious error', etc. does not allow one to TEST the veracity of your claim. Please tell us with specificity the study(s) and or researcher(s) you mean here.

Give explicit evidence of them trying to avoid statistics and probability (where appropriate).  I never saw any quoted study that overtly attempted to avoid the proper employment of analytic methods. When we restate a precise set of past empirically measured data, for instance, while one can apply inferential statistics, that may well be out of scope for a simple description of the actual data observed. For instance, any data set will exhibit an absolute upper and lower bound. There is no need to sum, square, sigma, percentile, or otherwise CALCULATE anything. What was; was.

#2. Once you identify the studies as requested above, I think you will find that each of them set out with a defined purpose. Many (most? All?) of the time, the objective was merely to describe in finite terms WHAT HAPPENED IN THE PAST, given certain variables, such as entry time, allocation, etc. Certainly the next step of someone else, using that information, might well be to seek an optimal strategy going forward, but I think you have the cart not only before the horse, but way over on top of the roof of the barn, when you claim what you do in the next clause, for a study like, say, Trinity...
 

Once you do that, please state (with specificity) what the ERROR is with the specifically identified study, within it's own stated bounds of purpose. Once you do this, I will happily stand with you to vigorously try to get any such errors admitted and corrected. That is the right thing to do, isn't it?

#3 Well, here we go. Unless you know the mind, scope and intent of such "Early Researchers" (which you will precisely identify, of course), how can a limitation on what the elect to investigate be in and of itself construed as an 'error', sir? I put it to you that it simply can not, on it's face. Scope, sir. Scope.

Think of it this way:  If I were to say that all of the research on your website is fatally flawed because you clearly failed to ever determine the best barbecue sauce, even after all that writing, and linking, and time -- you still simply and completely failed to ever say what the best BBQ sauce was, and therefore, you clearly wasted your time and my time and the website is useless. That would be silly, wouldn't it? Of course it would.
 
#4 This statement is meaningless without context. 'Valuations' are more important than... life? Oxygen? adequate funds to invest? Health? Return on the asset class? Unless you rework this to provide meaning and context it is, currently, simply without meaning. 

#5 Scope, again. Definition and purpose of the specific study will determine if this is of any import whatsoever. Many times, variables are modeled at the first order approximation, or ignored altogether if a researcher can make a determination they are likely not critical to the findings. Plainly, if we do not limit our need to control number of variables, we will forever be in a pool of changing conditions... lost peering into eddies of water, leafs, cloud, sun, birds on the horizon, frozen & apparently unable to start our controlled journey, when all we needed to do was note a strong continuous breeze for our sails, hoist, and begin...

#6-#11 are all similarly not ripe for discussion, as in #1-#5 until you have explicated to which studies you mean, and the specific issues you have with them. 

Please do not misunderstand the fact I stopped at #5 -- I am ENORMOUSLY willing to undertake dialog on this topic, but we must first be precise, in order to understand one another. No arm waving and generalities, if you please. 

DG