At the calculation stage, the SWR is a data-based construct.

Financial Independence/Retire Early -- Learn How!
Post Reply
User avatar
ataloss
**** Heavy Hitter
Posts: 559
Joined: Mon Nov 25, 2002 3:00 am

At the calculation stage, the SWR is a data-based construct.

Post by ataloss » Sat Nov 01, 2003 2:20 am

Calculation of the SWR is an objective exercise.
hocus
That's not what gummy, an ex-math professor, says.
wanderer

either hocus is confused about how swr is determined (say using monte carlo) or he is saying something meaningless

obviously with monte carlo, the inputs are best estimates ie non-objective
the output is thus an estimate (hence raddr's weasel words - in hocus' terminology)

so the swr is not objective, one can say that the claculation itself is objective although this is sort of irrelevant to the reliability of the swr
Have fun.

Ataloss

raddr
*** Veteran
Posts: 265
Joined: Mon Nov 25, 2002 3:25 am
Contact:

Re: At the calculation stage, the SWR is a data-based constr

Post by raddr » Sat Nov 01, 2003 4:14 am

ataloss wrote: obviously with monte carlo, the inputs are best estimates ie non-objective the output is thus an estimate (hence raddr's weasel words - in hocus' terminology)


Monte Carlo analysis is highly dependent on the inputs so, yes, unless you can predict the future then it is GIGO (garbage in, garbage out). Unless one possesses a reliable crystal ball then there will always be some uncertainty about the SWR going forward. That's why I prefer to give my SWR estimates in terms of probabilities in the context of making my assumptions perfectly clear. The latter is important since a single number or range of probabilities means nothing to me unless I know what assumptions are being made.
Last edited by raddr on Sat Nov 01, 2003 8:52 am, edited 1 time in total.

User avatar
ataloss
**** Heavy Hitter
Posts: 559
Joined: Mon Nov 25, 2002 3:00 am

Post by ataloss » Sat Nov 01, 2003 4:28 am

exactly, I think hocus is well aware of this (it has been discussed at length) but he prefers to be obscure- say that the mathematical operation is objective in a way that people can misinterpret this to mean that you think the swr is an objective "truth"
Have fun.

Ataloss

therealchips
*** Veteran
Posts: 174
Joined: Sat Jan 04, 2003 4:00 am
Location: Henderson, Nevada, USA

Do we agree at least on what SWR's have been historically?

Post by therealchips » Sat Nov 01, 2003 9:55 am

I am confused about the messages here. Don't we all agree at least that the historically calculated SWR's are objective and correct as history? That is, does anyone challenge the accuracy of the historical reporting in http://rehphome.tripod.com/pestudy1.html? Don't we all agree that in those charts, whether they use Shiller's Price/Earnings ratio based on current earnings or trailing 10-Year average earnings, the historically observed safe withdrawal rates have ranged from at least 4% to at most 11%? Correct me please if I am wrong, but so far as I know there is no quarrel as to what history shows on these matters. The quarrels have to do only with predicting future SWRs, don't they? Has anyone backtested his method of predicting future SWRs against the data? (I may very well have missed it. If the assertion is that present or Year 2000 conditions have never occurred in the data base, isn't backtesting impossible? Are people projecting the regression lines for SWR vs. PE into areas of PE never observed, but not recognizing the hazards of that projection?) Considering the paucity of the data, that is, how few thirty year periods are in the historical data, would that backtesting settle the issue of how to predict SWRs?

You know I have my reservations about the entire SWR enterprise anyway, since it omits so many considerations that are important to me, and I don't enjoy quarreling about it. For my purposes, I am satisfied in testing my withdrawal plans against the worst-case scenario the ataloss mentions. Anything worse that the worst ever observed is necessarily conjectural, by which I mean "permissible, maybe useful, but not convincing or conclusive".

Couldn't we at least agree on what history objectively shows before we disagree on how to make predictions?

<Wondering whether I should hit "delete" or "submit" . . . >
He who has lived obscurely and quietly has lived well. [Latin: Bene qui latuit, bene vixit.]

Chips

peteyperson
**** Heavy Hitter
Posts: 525
Joined: Tue Nov 26, 2002 6:46 am

Re: Do we agree at least on what SWR's have been historicall

Post by peteyperson » Sat Nov 01, 2003 1:40 pm

I, for one, am pleased you hit submit.

A typically good Chips post.

Petey
therealchips wrote: I am confused about the messages here. Don't we all agree at least that the historically calculated SWR's are objective and correct as history? That is, does anyone challenge the accuracy of the historical reporting in http://rehphome.tripod.com/pestudy1.html? Don't we all agree that in those charts, whether they use Shiller's Price/Earnings ratio based on current earnings or trailing 10-Year average earnings, the historically observed safe withdrawal rates have ranged from at least 4% to at most 11%? Correct me please if I am wrong, but so far as I know there is no quarrel as to what history shows on these matters. The quarrels have to do only with predicting future SWRs, don't they? Has anyone backtested his method of predicting future SWRs against the data? (I may very well have missed it. If the assertion is that present or Year 2000 conditions have never occurred in the data base, isn't backtesting impossible? Are people projecting the regression lines for SWR vs. PE into areas of PE never observed, but not recognizing the hazards of that projection?) Considering the paucity of the data, that is, how few thirty year periods are in the historical data, would that backtesting settle the issue of how to predict SWRs?

You know I have my reservations about the entire SWR enterprise anyway, since it omits so many considerations that are important to me, and I don't enjoy quarreling about it. For my purposes, I am satisfied in testing my withdrawal plans against the worst-case scenario the ataloss mentions. Anything worse that the worst ever observed is necessarily conjectural, by which I mean "permissible, maybe useful, but not convincing or conclusive".

Couldn't we at least agree on what history objectively shows before we disagree on how to make predictions?

<Wondering whether I should hit "delete" or "submit" . . . >

raddr
*** Veteran
Posts: 265
Joined: Mon Nov 25, 2002 3:25 am
Contact:

Re: Do we agree at least on what SWR's have been historicall

Post by raddr » Sat Nov 01, 2003 2:45 pm

therealchips wrote: Don't we all agree that in those charts, whether they use Shiller's Price/Earnings ratio based on current earnings or trailing 10-Year average earnings, the historically observed safe withdrawal rates have ranged from at least 4% to at most 11%?


I think that stock returns would've been considerably lower if we'd had modern-day liquidity and low cost index fund options available since 1871. For most of that period stocks were illiquid and very expensive to trade. Returns had to be higher to compensate for these factors but the historical studies assume liquid, low cost markets. The S&P500 didn't even exist for most of that time so how could one derive an SWR from it?

So no, I don't take the historical analysis at face value but YMMV. :wink:

peteyperson
**** Heavy Hitter
Posts: 525
Joined: Tue Nov 26, 2002 6:46 am

Re: Do we agree at least on what SWR's have been historicall

Post by peteyperson » Sat Nov 01, 2003 4:16 pm

Interesting points, raddr.

In which case, what sort of returns do you use for calculating a w/d rate you use?

I'm personally using 7% nominal for US, UK & Euope markets, 4% elsewhere, 7% REITS and 5.5% bonds. Sadly that drops the w/d rate down to circa 2.5%.

Petey
raddr wrote:
therealchips wrote: Don't we all agree that in those charts, whether they use Shiller's Price/Earnings ratio based on current earnings or trailing 10-Year average earnings, the historically observed safe withdrawal rates have ranged from at least 4% to at most 11%?


I think that stock returns would've been considerably lower if we'd had modern-day liquidity and low cost index fund options available since 1871. For most of that period stocks were illiquid and very expensive to trade. Returns had to be higher to compensate for these factors but the historical studies assume liquid, low cost markets. The S&P500 didn't even exist for most of that time so how could one derive an SWR from it?

So no, I don't take the historical analysis at face value but YMMV. :wink:

raddr
*** Veteran
Posts: 265
Joined: Mon Nov 25, 2002 3:25 am
Contact:

Re: Do we agree at least on what SWR's have been historicall

Post by raddr » Sat Nov 01, 2003 5:27 pm

peteyperson wrote: Interesting points, raddr.

In which case, what sort of returns do you use for calculating a w/d rate you use?


Petey,

For the S&P500 I'd postulate about a 3% real return based on the Gordon equation and other valuation parameters. For REITS I take the yield minus one. IOW if REITS are yielding 5% then I'd look for about a 4% real return. For other asset classes I don't really have any good rules to follow. I think that small caps, intl, gold etc. are not too far from fair value so I'd go pretty much with historical returns. These are really all just rough guesses but I think it pays to look at valuations when planning ahead. It certainly has served me well. :D

Post Reply