The Decumulation Phase

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
User avatar
grayfox
Posts: 4604
Joined: Sat Sep 15, 2007 4:30 am
Location: Earth

Re: The Decumulation Phase

Post by grayfox » Wed Aug 26, 2009 8:56 am

speedbump101 wrote:Today's paper had a write up on 'Unveiling the Retirement Myth' and mentioned it was available for free download (until Aug 31st) ... I took advantage of the invitation and found it quite interesting... It's particularily well suited for those approaching the draw down hase of their investment life cycle... Nothing earth shatteringly new, however I feel comfortable recommending it... If you want a quick overview of his strategy read chapter 41 first 'The Zone Strategy'

BTW to download the book in pdf form (non printable) he does require a name and email address.. Also note this is a Canadian author so there may be a slight difference in some of the terminology...

Paper article here: http://www.financialpost.com/opinion/st ... 98736079e8

Download the green 'pre-release' version here http://retirementoptimizer.com/

SB...


I have been reading Unveiling the Retirement Myth.

Finally somebody takes an intelligent look at the Distribution Phase based on worst case analysis. Up until now the only thing I have ever seen was variations on the Trinity study withdrawal method with floors, ceilings and buckets which are all nothing more than rearranging the deck chairs on the Titanic, in my opinion.

In a nutshell, what matters is your withdrawal rate and luck. What everybody here focuses on, asset allocation, is a distant third.

You are either in the green zone where you will succeed no matter what AA you have. The red zone, in which case you are doomed and no AA can help you. Or in the gray zone where you either get lucky or not.

livesoft
Posts: 57191
Joined: Thu Mar 01, 2007 8:00 pm

Post by livesoft » Wed Aug 26, 2009 9:12 am

I don't think this is a "finally", but more a once again. The FIRECalc retirement calculator appears to be identical to Otar's aftcast and it has been available for years and years. See it at www.firecalc.com FIREcalc is what most people over at the Early Retirement forum use to help them decide "Do I have enough?" and "How much can I withdraw annually?"

And guess what? The results are -- as expected -- essentially identical.

To be fair, I don't know which came first since both Otar and FIRECalc have been around a long, long time.

And both the Trinity study and Otar/URM come to the same conclusion: the SWR is about 4% (including taxes and investment expenses).

Otar's ideas may have been radical 10 years ago when he first started writing some of these articles that ended up in his book, but they are basically main stream nowadays.

In any event, Otar/URM is an outstanding contribution to your bookshelf. Even though I have the download, I will probably buy a hardcopy as well.

User avatar
grayfox
Posts: 4604
Joined: Sat Sep 15, 2007 4:30 am
Location: Earth

Post by grayfox » Wed Aug 26, 2009 9:39 am

livesoft wrote:
And both the Trinity study and Otar/URM come to the same conclusion: the SWR is about 4% (including taxes and investment expenses).


Otar seems to be both more pessimistic and conservative than Trinity and similar studies.

The Trinity study said 4% and many people interpret their charts to imply that a large equity allocation is necessary, like 60% or 75% equities.

By contrast, from Otar's chart on page 444, for a USA male aged 55 the green zone is below 3% and the red zone is above 4%. Female is even lower.

Also from Chap 16, his equity/fixed mix is the lessor of the Optimum AA and Tolerable AA, ends up more at 40 or 35 percent from simplified guidelines on page 175.

So at 4% where Trinity "safe" and conventional wisdom would have you at 75% stock,
4% is the start Otar's red zone where he would have you purchase annuities and only 40% stock or less.

livesoft
Posts: 57191
Joined: Thu Mar 01, 2007 8:00 pm

Post by livesoft » Wed Aug 26, 2009 10:00 am

The Trinity study was for a 30-year retirement nominally age 65 to age 95. Otar's Figure 41-5 on p. 444 gives 3.7% for the top of the Green zone for the same age range. To me 3.7% is "about 4%".

In the US, social security benefits are a CPI-linked annuity. These are included in FIREcalc if you so desire, so one could consider FIREcalc a kind of Gray Zone calculation which also helps to lead to about a 4% SWR.

As for 60% or 75% equities, we learn from Otar and others that you get about the same results if you are in the Green Zone over a broad range of equity percentages.

Anyways, everyone with a portfolio earmarked for retirement should read this book. If you like risk (I seem to be riskier than others on this forum), then you will land in his "hope" category. If you don't like risk, you will land in his "fear" category. There's something for everyone!

User avatar
HueyLD
Posts: 5164
Joined: Mon Jan 14, 2008 10:30 am

Post by HueyLD » Wed Aug 26, 2009 10:09 am

..............
Last edited by HueyLD on Sun Feb 08, 2015 7:18 am, edited 1 time in total.

User avatar
speedbump101
Posts: 997
Joined: Thu Oct 18, 2007 10:54 pm
Location: Alberta Canada

Post by speedbump101 » Wed Aug 26, 2009 10:18 am

livesoft wrote:The Trinity study was for a 30-year retirement nominally age 65 to age 95. Otar's Figure 41-5 on p. 444 gives 3.7% for the top of the Green zone for the same age range. To me 3.7% is "about 4%".



You just set off an alarm bell in my brain... Somewhere he cautions against doing exactly what you just did here. He suggests the consequences of small mental calculations having huge unanticipated implications ... Damn if I can find it now, however I'm sure someone in the crowd can help me.

SB...
"Man is not a rational animal, he is a rationalizing animal" -Robert A. Heinlein

livesoft
Posts: 57191
Joined: Thu Mar 01, 2007 8:00 pm

Post by livesoft » Wed Aug 26, 2009 10:30 am

Yes, 3.7% is different from 4.0%. But add in SS benefits and guess what?

I really like the way he starts out with failure mode analysis of a 6% withdrawal rate. Since when has anybody sanctioned a 6% withdrawal rate? Well, not since the Trinity et al. studies. :)

User avatar
Hexdump
Posts: 1610
Joined: Fri Oct 10, 2008 7:28 am
Location: Houston, Texas

I was a little surprised

Post by Hexdump » Wed Aug 26, 2009 10:56 am

at tables 17.1 and 17.2.

According to them, if your time horizon is 30+ years there is no need for TIPs in your portfolio, in order to maintain an acceptable sustainable withdrawal rate.

If I were to apply his percentages for our portfolio:
Time horizon = 40 years
SWR = 3.1 %
Equities = 42%
Money Markets = 6%
Bonds = 52%
TIPs = 0 % <----surprised me

Somehow I need to work this into our serious aversion to risk. Our ideal portfolio will be no more than 25% equities.

User avatar
grayfox
Posts: 4604
Joined: Sat Sep 15, 2007 4:30 am
Location: Earth

Post by grayfox » Wed Aug 26, 2009 1:02 pm

livesoft wrote:Yes, 3.7% is different from 4.0%. But add in SS benefits and guess what?

I really like the way he starts out with failure mode analysis of a 6% withdrawal rate. Since when has anybody sanctioned a 6% withdrawal rate? Well, not since the Trinity et al. studies. :)


One guy named Kitces, if I recall correctly, wrote a paper that takes valuation into account and now recommends a 6% w/d rate. Or at least did a few months ago.

User avatar
HueyLD
Posts: 5164
Joined: Mon Jan 14, 2008 10:30 am

Post by HueyLD » Wed Aug 26, 2009 4:24 pm

.............
Last edited by HueyLD on Sun Feb 08, 2015 7:18 am, edited 1 time in total.

User avatar
speedbump101
Posts: 997
Joined: Thu Oct 18, 2007 10:54 pm
Location: Alberta Canada

Post by speedbump101 » Wed Aug 26, 2009 5:37 pm

speedbump101 wrote:
livesoft wrote:The Trinity study was for a 30-year retirement nominally age 65 to age 95. Otar's Figure 41-5 on p. 444 gives 3.7% for the top of the Green zone for the same age range. To me 3.7% is "about 4%".


You just set off an alarm bell in my brain...

SB...


HueyLD wrote:Maybe this paragraph (pages 177-178) is what you could not find?

Withdrawal Rate:

In a distribution portfolio, the withdrawal rate is the most important contributor of portfolio longevity. It is far more important than asset allocation, asset selection, management fees, dividends, and reverse dollar cost averaging.

A seemingly small increase in withdrawals can change the outcome drastically. A case in point: If you have savings of $1 million at age 65 and your annual withdrawals are $38,000 indexed to inflation, market history shows that the probability of running out of money is only 9% by age 95. If you increase annual withdrawals by $4,000 to $42,000, then the probability of going broke by age 95 increases from 9% to 29%. Suddenly, an acceptable risk becomes a totally unacceptable one. You can play with any of the other variables such as asset allocation and management fees, and this will not likely change the picture.


Close Huey, but I think this is what set off the alarm.... p 188:

"Do not round up the sustainable withdrawal rate to an even 4%. We have seen earlier in the "Mathematics of Loss" (page 96) that for Bob 111's portfolio, a difference of 0.3% can make a difference between 2 million or nothing over 21 years." (my emphasis)

SB...
"Man is not a rational animal, he is a rationalizing animal" -Robert A. Heinlein

User avatar
bob90245
Posts: 6511
Joined: Mon Feb 19, 2007 8:51 pm

Post by bob90245 » Wed Aug 26, 2009 6:44 pm

speedbump101 wrote:
speedbump101 wrote:
livesoft wrote:The Trinity study was for a 30-year retirement nominally age 65 to age 95. Otar's Figure 41-5 on p. 444 gives 3.7% for the top of the Green zone for the same age range. To me 3.7% is "about 4%".


You just set off an alarm bell in my brain...

SB...


HueyLD wrote:Maybe this paragraph (pages 177-178) is what you could not find?

Withdrawal Rate:

In a distribution portfolio, the withdrawal rate is the most important contributor of portfolio longevity. It is far more important than asset allocation, asset selection, management fees, dividends, and reverse dollar cost averaging.

A seemingly small increase in withdrawals can change the outcome drastically. A case in point: If you have savings of $1 million at age 65 and your annual withdrawals are $38,000 indexed to inflation, market history shows that the probability of running out of money is only 9% by age 95. If you increase annual withdrawals by $4,000 to $42,000, then the probability of going broke by age 95 increases from 9% to 29%. Suddenly, an acceptable risk becomes a totally unacceptable one. You can play with any of the other variables such as asset allocation and management fees, and this will not likely change the picture.


Close Huey, but I think this is what set off the alarm.... p 188:

"Do not round up the sustainable withdrawal rate to an even 4%. We have seen earlier in the "Mathematics of Loss" (page 96) that for Bob 111's portfolio, a difference of 0.3% can make a difference between 2 million or nothing over 21 years." (my emphasis)

SB...

Geez! I am very surprised people are fussing over trivial things like 0.3%. Unlike physical properties, investing is a VERY inexact science. The precision one is assuming here simply does not exist. There are so many external variables that can tilt the assumed probabilities one way or the other.

Come on, people! 4% SWR is only a rough rule of thumb. Not a ruler that can measure precisely to the nearest tenth of an inch.

:roll:
Ignore the market noise. Keep to your rebalancing schedule whether that is semi-annual, annual or trigger bands.

User avatar
speedbump101
Posts: 997
Joined: Thu Oct 18, 2007 10:54 pm
Location: Alberta Canada

Post by speedbump101 » Wed Aug 26, 2009 7:59 pm

bob90245 wrote:Come on, people! 4% SWR is only a rough rule of thumb. Not a ruler that can measure precisely to the nearest tenth of an inch.

:roll:


I'm just quoting what the author wrote... He made a 'specific' point of saying don't do it... I agree that this isn't an exact science however since we are discussing his book here and he says specifically not to do this because of the reasons mentioned I think it's topical... Coincidentally the rounding LS applied happens exactly the same numbers the author used on page 96... In the author's example this .3% had a huge effect. Have you read his chapter on this?

The nit picking charge excepted, I'd say this is a very interesting and well researched book... Well worth the effort required to digest its 525 pages... We may not agree on all his points, however I'd be surprised if anyone reading it doesn't admit to learning something from Otar's approach to financing his / her retirement years.

SB...
Last edited by speedbump101 on Wed Aug 26, 2009 11:22 pm, edited 2 times in total.
"Man is not a rational animal, he is a rationalizing animal" -Robert A. Heinlein

Willy
Posts: 78
Joined: Mon Jul 09, 2007 8:14 pm

Post by Willy » Wed Aug 26, 2009 8:26 pm

I found Otar's book to be interesting and helpful. It left me wondering two things, though:

1. If one assumes expense ratios of 25 basis points (as is common for "Bogleheads") instead 200 basis points, how much would the safe withdrawal percentage be increased?

2. What is the best way to apply his zone analysis in the common situation where one's withdrawal rate will initially be on the high side, before dropping well into the safe zone once Social Security benefits kick in?

Maybe Otar addresses these issues and I missed it by reading too quickly. If he doesn't, what do others think?

User avatar
bob90245
Posts: 6511
Joined: Mon Feb 19, 2007 8:51 pm

Post by bob90245 » Wed Aug 26, 2009 8:26 pm

No, I haven't read the author's book. No doubt, he has some good thoughts. But this over-preoccupation with precision is not one of them.
Ignore the market noise. Keep to your rebalancing schedule whether that is semi-annual, annual or trigger bands.

User avatar
Sheepdog
Posts: 4791
Joined: Tue Feb 27, 2007 3:05 pm
Location: Indiana, retired 1998 at age 65

Post by Sheepdog » Wed Aug 26, 2009 8:30 pm

Here I am again. Don't adjust for inflation. Set your withdrawal percentage (I recommend in the 4 to 5% range, but it doesn't have to be) based on your current portfolio balance. Let your investment growth be your inflation hedge. Have more, you can spend more and vice versa. You will never run out. Just have an investment portfolio in your comfort zone when you start out. Caveat: Don't always spend your annual maximum goal, though. Try to budget to be below that amount and save your unspent dollars in your "budget bank" for the year when you need to spend more than your goal, like your round the world cruise or your new car.
Jim
People should not say everything they think. They should think about everything they say.

User avatar
speedbump101
Posts: 997
Joined: Thu Oct 18, 2007 10:54 pm
Location: Alberta Canada

Post by speedbump101 » Wed Aug 26, 2009 8:55 pm

Willy wrote:I found Otar's book to be interesting and helpful. It left me wondering two things, though:

1. If one assumes expense ratios of 25 basis points (as is common for "Bogleheads") instead 200 basis points, how much would the safe withdrawal percentage be increased?

2. What is the best way to apply his zone analysis in the common situation where one's withdrawal rate will initially be on the high side, before dropping well into the safe zone once Social Security benefits kick in?

Maybe Otar addresses these issues and I missed it by reading too quickly. If he doesn't, what do others think?



Re 2) see p337 re Immediate term certain annuities... I don't know for sure what Mr. Otar would suggest, however this seems to be a likely suggestion for someone out of the green zone, and wishing to out source the specific time period risk you describe.... Maybe we could coax Mr. Otar back into this discussion.

SB...
"Man is not a rational animal, he is a rationalizing animal" -Robert A. Heinlein

livesoft
Posts: 57191
Joined: Thu Mar 01, 2007 8:00 pm

Post by livesoft » Wed Aug 26, 2009 9:10 pm

Willy, check out www.firecalc.com for ideas on adding SS at time points in the future and adjusting expense ratios. FIREcalc also has a few spending models as well, so that if your portfolio drops in value, you can model a smaller withdrawal for those cases as well.

livesoft
Posts: 57191
Joined: Thu Mar 01, 2007 8:00 pm

Post by livesoft » Fri Aug 28, 2009 7:54 pm

grayfox wrote:
livesoft wrote:Yes, 3.7% is different from 4.0%. But add in SS benefits and guess what?

I really like the way he starts out with failure mode analysis of a 6% withdrawal rate. Since when has anybody sanctioned a 6% withdrawal rate? Well, not since the Trinity et al. studies. :)


One guy named Kitces, if I recall correctly, wrote a paper that takes valuation into account and now recommends a 6% w/d rate. Or at least did a few months ago.


And today's paper has Kitces being discussed:
http://www.nytimes.com/2009/08/29/your- ... money.html

dbr
Posts: 24123
Joined: Sun Mar 04, 2007 9:50 am

Re: The Decumulation Phase

Post by dbr » Sat Aug 29, 2009 8:26 am

bob90245 wrote:
cotar@rogers.com wrote:If your required withdrawals (required by the retiree, not the MRD) is less than the sustainable withdrawal rate, then your portfolio generally continues to increase in value. However, the RMD forces to take you out more than what you need. So, the growth of your portfolio is significantly reduced, whether you are lucky, median or unlucky (Please see charts in Figure 20.5) and you wanted to withdraw only a small amount (less than SWR), compared to if no RMD were to apply to that portfolio.

Correct me if I'm wrong, but the retiree is not force to spend all of the RMD. So part can pay the taxes, part can fund current expenses and part can be invested in an taxable account. Yes?



I'm bumping this back up as Mr. Otar didn't answer this question and hoping he is still here and will look at it. As pointed out by Bob here, there has to be a subtlety that is escaping us.

conundrum
Posts: 817
Joined: Sat May 09, 2009 7:00 pm

Post by conundrum » Sat Aug 29, 2009 7:31 pm

Just a comment, not on the book but on Jim Otar. I have been diligently trying to download the $3.99 copy and have had all sorts of problems doing so. Jim has been very helpful and has returned all my e-mails very promptly. I unfortunately cannot yet comment on the book but can say that Jim has been very helpful.

Drum

User avatar
speedbump101
Posts: 997
Joined: Thu Oct 18, 2007 10:54 pm
Location: Alberta Canada

Post by speedbump101 » Sun Aug 30, 2009 10:21 am

Here is Jim Otar's second video interview...

Slow to load, have patience.

http://www.financialpost.com/video/inde ... BjdJAMlKDS

IMO I believe we'd be privileged if Jim continues to participate on this forum... He has spent the better part of six years amassing the data for this book, and having his knowledge available (occasionally) to us here would be outstanding.

SB...
"Man is not a rational animal, he is a rationalizing animal" -Robert A. Heinlein

User avatar
Hexdump
Posts: 1610
Joined: Fri Oct 10, 2008 7:28 am
Location: Houston, Texas

Hear, hear.

Post by Hexdump » Sun Aug 30, 2009 10:31 am

speedbump101 wrote:Here is Jim Otar's second video interview...

Slow to load, have patience.

http://www.financialpost.com/video/inde ... BjdJAMlKDS

IMO I believe we'd be privileged if Jim continues to participate on this forum... He has spent the better part of six years amassing the data for this book, and having his knowledge available (occasionally) to us here would be outstanding.

SB...


I will host a few pints of Molson's Golden to that.

cjking
Posts: 1750
Joined: Mon Jun 30, 2008 4:30 am

Post by cjking » Sun Aug 30, 2009 10:50 am

conundrum wrote:Just a comment, not on the book but on Jim Otar. I have been diligently trying to download the $3.99 copy and have had all sorts of problems doing so. Jim has been very helpful and has returned all my e-mails very promptly. I unfortunately cannot yet comment on the book but can say that Jim has been very helpful.

Drum


I emailed him with a question prompted by the book, and also got a very prompt and useful answer.

I think this book is going to be a valuable resource for investors generally. Investing forum addicts may know about "sequence of returns", this book will bring issues like that to a wider audience.

I particularly liked his tests for a failing retirement. I can understand that people retiring during the equity bubble might have been reluctant or unable to confront equity valuation issues, his tests would have warned them they had a problem without requiring them to take on board a whole body of knowledge, which conventional wisdom didn't regard as worth worrying about. (And still doesn't.)

cjking
Posts: 1750
Joined: Mon Jun 30, 2008 4:30 am

Post by cjking » Sun Aug 30, 2009 10:58 am

livesoft wrote:I don't think this is a "finally", but more a once again.


The fact that asset allocation and diversification are actually relatively insignificant factors in retirement success must surely be a suprise to some Bogleheads?

These are stock answers whenever questions about risk control arise, though to be fair those question aren't always in the context of retirement.

dbr
Posts: 24123
Joined: Sun Mar 04, 2007 9:50 am

Post by dbr » Sun Aug 30, 2009 12:00 pm

cjking wrote:
livesoft wrote:I don't think this is a "finally", but more a once again.


The fact that asset allocation and diversification are actually relatively insignificant factors in retirement success must surely be a suprise to some Bogleheads?

These are stock answers whenever questions about risk control arise, though to be fair those question aren't always in the context of retirement.


As an investment forum, Bogleheads does not particularly focus on how to fund retirement. Risk control in investing is indeed a subject that is completely different from risk control in retirement funding.

User avatar
iceport
Posts: 3333
Joined: Sat Apr 07, 2007 4:29 pm

Post by iceport » Sun Aug 30, 2009 1:13 pm

cjking wrote:
livesoft wrote:I don't think this is a "finally", but more a once again.

The fact that asset allocation and diversification are actually relatively insignificant factors in retirement success must surely be a suprise to some Bogleheads?

These are stock answers whenever questions about risk control arise, though to be fair those question aren't always in the context of retirement.

Well, yes and no. (I've finished up to Ch. 5, so I guess I can comment on this?) That the basic stock/bond ratio is insignificant is indeed a surprise, and as grayfox noted does not seem entirely consistent with the Trinity study. However, the diversification chapter is generally more Bogle-like than many Bogle-heads seem to be. Specifically, Bogle and Otar agree that the quest for elaborate diversification schemes within equities is not too important (think TSM "lumper") and that the maximum international allocation should be something like 20%. Both of those ideas are vehemently opposed by many here. On the other hand, Otar, Bogle, and most of the forum gurus all seem to agree that there's no way of knowing which portfolio will perform the best going forward.

All also seem to agree that diversification across stocks and bonds is important (Otar in Ch. 5 conclusion).

One big divergence from the Boglehead EMH inspired belief in indexing is Otar's conclusion that "finding a few good portfolio managers and hanging onto them as long as they continue to perform well" is better than trying to be more broadly diversified. So much for "buy and hold the market".

As little as I've read, I do find the original thinking and unconventional approach helpful and refreshing, even as some of it leaves me scratching my head.

--Pete

dbr
Posts: 24123
Joined: Sun Mar 04, 2007 9:50 am

Post by dbr » Sun Aug 30, 2009 1:37 pm

I don't think Otar's thinking is hugely original and unconventional within the context of the last decade or so in looking at this problem. His overall insight is pure Trinity in its essence. The work in early retirement culminating in the development of FireCalc anticipates most of what Otar is writing about. I think Otar does an excellent job underlining the key points and of placing the analysis in a correct perspective. A weakness might be that the approach attempts to be more quantitative than the data might allow.

The relative insensitivity of the retirement funding problem to AA compared to the emphasis on AA in an investing forum is a disconnect that is/has been mentioned here periodically. Similarly the issue of sequence of returns and luck features large in retirement funding and is pretty much out of context as far as "how to invest."

User avatar
speedbump101
Posts: 997
Joined: Thu Oct 18, 2007 10:54 pm
Location: Alberta Canada

Post by speedbump101 » Mon Aug 31, 2009 7:27 pm

dbr wrote:I don't think Otar's thinking is hugely original and unconventional within the context of the last decade or so in looking at this problem. His overall insight is pure Trinity in its essence. The work in early retirement culminating in the development of FireCalc anticipates most of what Otar is writing about. I think Otar does an excellent job underlining the key points and of placing the analysis in a correct perspective. A weakness might be that the approach attempts to be more quantitative than the data might allow.

The relative insensitivity of the retirement funding problem to AA compared to the emphasis on AA in an investing forum is a disconnect that is/has been mentioned here periodically. Similarly the issue of sequence of returns and luck features large in retirement funding and is pretty much out of context as far as "how to invest."


"Trinity like in essence"... yes, but here is a difference that jumped off the page when I read it:

"Historical annual return data were used to calculate ending portfolio values after annual dollar withdrawals; the annual dollar withdrawals are based on a first-year withdrawal rate that is a percentage of the initial portfolio value. For instance, for a 100% stock portfolio with a 15-year payout and a 3% initial withdrawal rate, the amount remaining after the payout period was determined at the end of the first 15-year period (1926 to 1940), the second 15-year period (1927 to 1941), etc. The portfolio success rate in the study is the percentage of all past payout periods supported by the portfolio (where the ending value exceeds $0). [For those more technically inclined, an illustration of the algorithm used can be found at the AAII Journal Web site at Algorithm."

http://bobsfiles.home.att.net/trinity.htm

Otar p181 talking about SWRs:

"Beware that many of the academic studies that talk about sustainable withdrawal rates are flawed because of one of the following reasons:

- Start–date bias: Many studies use 1926 as a starting year. This start–date treats the 1929 crash as a “one–time” event. Start in 1900 and you will see how the 1929 crash appears like a black hole."


I 'think' the Trinity rolling periods are the same as Otar's 'sequence of returns'... If not I'm sure I'll be corrected. :-)

SB...
"Man is not a rational animal, he is a rationalizing animal" -Robert A. Heinlein

User avatar
bob90245
Posts: 6511
Joined: Mon Feb 19, 2007 8:51 pm

Post by bob90245 » Mon Aug 31, 2009 7:50 pm

speedbump101 wrote:Otar p181 talking about SWRs:

"Beware that many of the academic studies that talk about sustainable withdrawal rates are flawed because of one of the following reasons:

- Start–date bias: Many studies use 1926 as a starting year. This start–date treats the 1929 crash as a “one–time” event. Start in 1900 and you will see how the 1929 crash appears like a black hole."

So instead use http://firecalc.com/ which uses 1871 as the starting year.
Ignore the market noise. Keep to your rebalancing schedule whether that is semi-annual, annual or trigger bands.

User avatar
speedbump101
Posts: 997
Joined: Thu Oct 18, 2007 10:54 pm
Location: Alberta Canada

Post by speedbump101 » Mon Aug 31, 2009 8:55 pm

bob90245 wrote:
speedbump101 wrote:Otar p181 talking about SWRs:

"Beware that many of the academic studies that talk about sustainable withdrawal rates are flawed because of one of the following reasons:

- Start–date bias: Many studies use 1926 as a starting year. This start–date treats the 1929 crash as a “one–time” event. Start in 1900 and you will see how the 1929 crash appears like a black hole."

So instead use http://firecalc.com/ which uses 1871 as the starting year.


Thanks... Ran a few and the results are pretty darn close... any difference might be explained by the fact that FI in Otar's (the data I chose) is 6Mo CD plus .05% whereas FC's shortest FI, using 1900 is 5 yr treasury.

90% confidence level in Otar's versus 1 failure in 80 cycles in FC... ($38000 from $1,000,000 as base AA 35/65 E/B, 30 yrs)

Just trying to calibrate the two data sources, I realize real life doesn't work this way.

SB...
"Man is not a rational animal, he is a rationalizing animal" -Robert A. Heinlein

dbr
Posts: 24123
Joined: Sun Mar 04, 2007 9:50 am

Post by dbr » Tue Sep 01, 2009 9:34 am

dbr wrote:I don't think Otar's thinking is hugely original and unconventional within the context of the last decade or so in looking at this problem. His overall insight is pure Trinity in its essence. The work in early retirement culminating in the development of FireCalc anticipates most of what Otar is writing about. I think Otar does an excellent job underlining the key points and of placing the analysis in a correct perspective. A weakness might be that the approach attempts to be more quantitative than the data might allow.

The relative insensitivity of the retirement funding problem to AA compared to the emphasis on AA in an investing forum is a disconnect that is/has been mentioned here periodically. Similarly the issue of sequence of returns and luck features large in retirement funding and is pretty much out of context as far as "how to invest."


Looking at what I wrote here, I would indeed like to add a big thank you to Mr. Otar for contributing such a well written discussion of how this problem works together with so much expanded analysis and the development of a tool to implement the thinking.

I still think it is appropriate for thinking to be done regarding how the analysis and the tool might somehow be mistaken or misleading in that attempting to knock down an approach is the best way to firmly establishing the soundness of an approach.

As of now, the only thing that jumps to mind is what appears on first impression to be an underlying assumption that the characteristics of US financial market over the last century or so are good enough data for planning the future few decades of someone retiring today. Naturally trying to suggest what might replace this is a massive challenge.

User avatar
HueyLD
Posts: 5164
Joined: Mon Jan 14, 2008 10:30 am

Re: The Decumulation Phase

Post by HueyLD » Tue Sep 01, 2009 1:44 pm

................
Last edited by HueyLD on Sun Feb 08, 2015 7:20 am, edited 1 time in total.

User avatar
bob90245
Posts: 6511
Joined: Mon Feb 19, 2007 8:51 pm

Post by bob90245 » Tue Sep 01, 2009 2:02 pm

OK so it is simply a tax-related phenomenon. The retiree has to share the wealth with the IRS. And this saps the growth generating potential of the assets. I see where he is coming from now.
Ignore the market noise. Keep to your rebalancing schedule whether that is semi-annual, annual or trigger bands.

User avatar
HueyLD
Posts: 5164
Joined: Mon Jan 14, 2008 10:30 am

Post by HueyLD » Tue Sep 01, 2009 2:16 pm

............
Last edited by HueyLD on Sun Feb 08, 2015 7:22 am, edited 1 time in total.

User avatar
iceport
Posts: 3333
Joined: Sat Apr 07, 2007 4:29 pm

Post by iceport » Tue Sep 01, 2009 10:59 pm

HueyLD wrote:
petrico wrote:One big divergence from the Boglehead EMH inspired belief in indexing is Otar's conclusion that "finding a few good portfolio managers and hanging onto them as long as they continue to perform well" is better than trying to be more broadly diversified. So much for "buy and hold the market".--Pete

To do that Pete, you will need to really dig into chapters 24 - 28. Pretty complicated mathematical stuff indeed. I am still trying to understand these. Indexing has worked relatively well for me, but I have an open mind for different ideas as long as they are supported by good data and analysis.

You are probably more open-minded than me! But even if I can't be persuaded to adopt active management -- or any other aspect of Otar's recommendations -- I will certainly try very hard not to let that detract from the rest of his message.

--Pete

JOtar
Posts: 21
Joined: Sun Aug 23, 2009 5:35 pm

The Decumulation Phase

Post by JOtar » Wed Sep 02, 2009 9:39 am

Very interesting comments above.

1. RMD question/comments. I think HueyLD summed it up extremely well. No need to add another word from me.

2. Trinity study, or any other studies: I ignore (and throw away) any academic study that: A. uses some kind of simulator, Monte Carlo or otherwise, B. uses "generous" historical dividends, C. data starting in 1926 or later.

Some may call it arrogant, but I find it mentally less cluttering, because it eliminates 98% of the academic research from my "inbox".

For example, I contributed a chapter on the luck factor to "Retirement Income Redesigned" a few years ago. Eventually, I received one complimentary copy of the book from the publisher. I could only find one chapter in it that was trustworthy to read, and that was W. Bengen's chapter. That is because he also uses actual market history. Many other chapters may be great reading in theory, but they used simulators and other approaches that have weak "legs" (see page 292, "Unveiling the Retirement Myth").

3. Otar Retirement Calculator: I developed it in year 2000 when I was writing my first book "High Expectations and False Dreams", where I talked about the adverse effects of the sequence of returns, inflation and reverse dollar cost averaging, basically the three components of the luck factor several years before many others jumped later on on the bandwagon.

At that time, when I published that book, I made a simpler retirement calculator available on my website for almost free (I asked people to donate to Cancer Society) that reflected the sequence of returns to some extent (2-layered Monte Carlo). A few fellow advisors asked me to make the Otar Retirement Calculator that I used for the first book, available for them in a more user friendly edition. At that time it was a 45 mB program which had everything from life insurance to owning your second home in it. So, I chopped away everything that is not related to retirement planning, reduced the size to 15 mB and made the program available for others since 2004. Yes, you can change any income at any age, higher lower, buy a car every three years, include investment portfolios, VA-GMWB, VA-GMIB, life annuity, VPA etc in your income allocation and see results.

Firecalc: If you are using Firecalc, then you are already miles ahead of the crowd. As long as you are using the actual history, a history that is not contaminated by statistics or infested by academic theories, you are on the right track. I don't know when Firecalc started, you should ask it to them. When I was looking at calculators in 2004, before I made mine commercially available, there was no calculator available then based on actual market history, as far as I could tell then.

4. Be safe, do not round up 3.7% to 4%. It is not worth going broke at age 91, when you might not know the difference between the urinal and the mattress. At that time, it is too late to make adjustments of 0.3%.

To the skeptics that say "can you rely on historical outcomes?" My answer is "No, I can't". I think the outcome will be worse in this century than last century. The culture of short-term thinking and unethical behavior is much more ingrained now then before and it is all around us. We are, in effect, stealing from our children by downloading all this debt and burden to them, just for us to feel good for a while longer. So, if you want to be even safer, keep your SWR to 3.3% to account for that. And if you have been living frugally, meeting this benchmark should not be that difficult.

have a wonderful day

Levett
Posts: 4177
Joined: Fri Feb 23, 2007 2:10 pm
Location: upper Midwest

Post by Levett » Wed Sep 02, 2009 10:52 am

Spot on, Mr. Otar. Would that others had your ability to cut to the chase. Bob U.

P.S. I, too, think HueyLDs response re RMD was excellent.
There are some things that count that can't be counted, and some things that can be counted that don't count.

dbr
Posts: 24123
Joined: Sun Mar 04, 2007 9:50 am

Post by dbr » Wed Sep 02, 2009 11:34 am

If the outcome of analysis using historical data is so sensitive to the time period chosen that data from 1871 on is valid (FireCalc), but data only from 1926 on is garbage, then there is a fundamental problem with the analysis. One could just as well assert that data only from 1871 is garbage but data from 1851 is valid, or that data including 1929 is so unlikely to be repeated that that time period should be excluded, etc., etc.

One thing that needs to be examined with any simulation or forecast is the sensitivity to the input data.

A different aspect of the same issue is that a forecast should always be presented with an estimate of the range of uncertainty that applies to the results. Answering to that challenge is one of the primary purposes in using statistical methods such as MC. Naturally if one uses MC and neglects publishing the uncertainties, then the job isn't really done.

As far as 3.7% vs. 4.0%, it is entirely plausible that retirement outcomes in the model are sensitive enough to withdrawal rate to warn against rounding off at that level. That is not the same question as asking if the model is capable of reflecting reality to an accuracy better than a tenth of a percent out of four percent. Certainly if the fudge for not being able to rely precisely on historical data is perhaps 0.4% then it is reasonable to consider 3.7% and 4.0% to be the same withdrawal rate for purposes of practical advice.

I would conclude by emphasizing, however, that it is far, far better to operate with a model that demonstrates approximately how things really work than to exercise false assumptions and flights of fantasy in retirement planning.

User avatar
speedbump101
Posts: 997
Joined: Thu Oct 18, 2007 10:54 pm
Location: Alberta Canada

Post by speedbump101 » Wed Sep 02, 2009 12:07 pm

dbr wrote:If the outcome of analysis using historical data is so sensitive to the time period chosen that data from 1871 on is valid (FireCalc), but data only from 1926 on is garbage, then there is a fundamental problem with the analysis. One could just as well assert that data only from 1871 is garbage but data from 1851 is valid, or that data including 1929 is so unlikely to be repeated that that time period should be excluded, etc., etc.

One thing that needs to be examined with any simulation or forecast is the sensitivity to the input data.

A different aspect of the same issue is that a forecast should always be presented with an estimate of the range of uncertainty that applies to the results. Answering to that challenge is one of the primary purposes in using statistical methods such as MC. Naturally if one uses MC and neglects publishing the uncertainties, then the job isn't really done.

As far as 3.7% vs. 4.0%, it is entirely plausible that retirement outcomes in the model are sensitive enough to withdrawal rate to warn against rounding off at that level. That is not the same question as asking if the model is capable of reflecting reality to an accuracy better than a tenth of a percent out of four percent. Certainly if the fudge for not being able to rely precisely on historical data is perhaps 0.4% then it is reasonable to consider 3.7% and 4.0% to be the same withdrawal rate for purposes of practical advice.

I would conclude by emphasizing, however, that it is far, far better to operate with a model that demonstrates approximately how things really work than to exercise false assumptions and flights of fantasy in retirement planning.


Ok I admit I'm outa my league here, however that little fact has never stopped me before :-).

dbr, I ‘think’ the point here is that we all know that the Great Depression has great significance from a historical data perspective... I also 'think' Jim's point is that this significance can be minimized more than he feels comfortable with by using data starting in 1926, which just happens to be what the Trinity study did...

When using rolling periods starting with 1926 to current, obviously the depression’s reversion below the mean affects fewer periods than using a 1871 start date... I ‘think’ that was his point.

Obviously you are much better versed in statistics than me, so please adjust your expectations accordingly :-)

SB...
"Man is not a rational animal, he is a rationalizing animal" -Robert A. Heinlein

dbr
Posts: 24123
Joined: Sun Mar 04, 2007 9:50 am

Post by dbr » Wed Sep 02, 2009 1:32 pm

There is no quarrel at all that more information about the behavior of the system is better than less information. The question (and it is a question, only) is with whether or not the extra 55 years make the difference between obtaining accurate projections for retirement planning and getting garbage. My point is not about statistical nuances but about estimating the accuracy of the model. One component of that is finding out the sensitivity to input data that may be selected in an arbitrary manner from among numerous choices. A concern that one can have (not saying one does have) is that historical data used to project the future needs to still be representative of the future system. It is worth studying which ranges of time best suite that criterion, although doing so may be difficult to accomplish.

The strength in Mr. Otar's approach, with which I agree completely, is that historic data includes the wild gyrations and long trends that actual investment history contains that naive (emphasis on naive) MC simulations fails to incorporate. Whether an 1871 data set or a 1926 data set incorporates too much, too little, or just right should be an interesting discussion and much deserving of study.

Here is a question for Mr. Otar: My sense is that your approach and the results are certainly in the spirit of what is revealed by Trinity and in fact not drastically different quantitatively from Trinity, within the limits of the data. Can you elucidate what is so flawed about their study within the context of their purpose, that you object so strongly to it?

I would also like to add that engaging in discussion of something is hardly the same thing as to be a "skeptic." Discussion is for the purpose of enhancing understanding and adding perspective, which is my objective in posting here.

I am also sorry that I will not be available on the forum for the next couple of weeks and may not be able to respond in this thread after today.

Thank you.

User avatar
dothemontecarlo
Posts: 532
Joined: Tue Nov 06, 2007 10:19 am
Location: Colorado
Contact:

Post by dothemontecarlo » Wed Sep 02, 2009 2:16 pm

dbr wrote:If the outcome of analysis using historical data is so sensitive to the time period chosen that data from 1871 on is valid (FireCalc), but data only from 1926 on is garbage, then there is a fundamental problem with the analysis. One could just as well assert that data only from 1871 is garbage but data from 1851 is valid, or that data including 1929 is so unlikely to be repeated that that time period should be excluded, etc., etc.


One problem with exploratory simulation in general -- unless you "loop" the returns and sample every unique interval in the "loop" -- is that you end up undersampling the data near the ends of your data set, like the graph below illustrates:

Image

(BTW: that's why TIP$TER loops the data for its exploratory simulation).

If your historial data sequence starts in 1926, and you don't loop your data, then you are going to undersample the Great Depression (relative to how much you sample data in the 1956-1979 period).

dbr
Posts: 24123
Joined: Sun Mar 04, 2007 9:50 am

Post by dbr » Wed Sep 02, 2009 2:33 pm

Hi Don, I was actually just looking at the situation you illustrate above on your page comparing TIP$TER and FireCalc, and it is a good observation.

Granting that it is always easier to ask someone who has done something rather than doing something oneself, I am curious if you have tabulated the amount of variation one gets in the results as a function of range of historical periods and the way the loop is done?

It seems that it is a reasonable question whether one should underweight, evenly weight, or overweight extraordinary periods in an analysis like this.

One of the concerns in using historical periods analysis is that specific significant departures from average in the past may not repeat, but that what will occur in the future that is extraordinary will be different from any example observed in the past. That is not to say that a non-historical model such as MC solves that problem at all.

Your contribution of TIP$TER, by the way, is much appreciated.

User avatar
dothemontecarlo
Posts: 532
Joined: Tue Nov 06, 2007 10:19 am
Location: Colorado
Contact:

Post by dothemontecarlo » Wed Sep 02, 2009 3:27 pm

dbr wrote:Hi Don, I was actually just looking at the situation you illustrate above on your page comparing TIP$TER and FireCalc, and it is a good observation.

Granting that it is always easier to ask someone who has done something rather than doing something oneself, I am curious if you have tabulated the amount of variation one gets in the results as a function of range of historical periods and the way the loop is done?


I haven't, and to be honest I haven't thought of a real good way to do that comparison. Since it might vary depending on the parameters you choose for a "test person."

Playing around with it myself, I get the "feeling" that the looping increases the failure rate somewhat, because when I click on the "Choose Your Percentile" option to see which simulated periods accounted for the 0, 5, 10, 15, 20 percentile, etc., outcomes, intervals that include the "spliced" part of the data set show up pretty frequently. That's probably because this decade, so far, has been relatively awful.

dbr wrote:One of the concerns in using historical periods analysis is that specific significant departures from average in the past may not repeat, but that what will occur in the future that is extraordinary will be different from any example observed in the past. That is not to say that a non-historical model such as MC solves that problem at all.


That's a very good insight. What if the 1871-2008 data set represents the top end of the yet-to-seen 1871-2100 bell curve? It could end up that way.

That's why I kept several MC simulation options, so one can try out different simulation models. (It's kind of interesting comparing them anyway).

dbr wrote:Your contribution of TIP$TER, by the way, is much appreciated.


Thanks! That means a lot. :D

BTW, in my spare time I'm working on making it faster (w/ a C++ DLL). The current ~four minutes (on Excel 2007) it takes to run an asset allocation "spectrum" simulation (i.e., 0% to 100% AAs, in 10% increments) is way too long. Goal: 15 seconds or less (30 seconds max). Wish me luck.

dbr
Posts: 24123
Joined: Sun Mar 04, 2007 9:50 am

Post by dbr » Wed Sep 02, 2009 3:33 pm

Good luck!

JOtar
Posts: 21
Joined: Sun Aug 23, 2009 5:35 pm

The Decumulation Phase

Post by JOtar » Sat Sep 05, 2009 6:29 pm

Feel free to use my calculator to optimize the asset mix based on actual market history since 1900, it takes about 10 seconds in total in my equipment. It includes all your cash flow in/out, probability of depletion, soonest age of portfolio depletion, median/unlucky portfolio values for a range of asset mixes between 0% and 100% equity content. The current age is fixed in the free trial version at 55, but it is close enough for a test run.

User avatar
speedbump101
Posts: 997
Joined: Thu Oct 18, 2007 10:54 pm
Location: Alberta Canada

Post by speedbump101 » Sat Sep 05, 2009 10:05 pm

I have been playing with the trial version of Otar's calculator and I'm very impressed with the scope of data that it can crunch... One heads up though, (for me anyway) is that the user guide button on the main page didn't work ... I had to find the file manually, and open it from the root directory... Its default location is C:\ORCTrial\UserGuide.pdf (132 pages in the guide... yes it's 'rather' comprehensive). I haven’t seen any other posters experiencing this problem so I may be an exception, however believe me you do need to read the manual to achieve full functionality.

Before I scare everyone off, it's worth mentioning that this calculator can be run in a very simple mode, or an extremely complex one, incorporating laddered annuities, periodic income... you name it, and Jim Otar seems to have included it... Heaven knows how many hours went into developing this software, but it sure wasn’t an overnight project. :-)

Edit: Macros are enabled and all other buttons work correctly... with the exception of the guide, and disclosures hyperlinks.

SB...
"Man is not a rational animal, he is a rationalizing animal" -Robert A. Heinlein

User avatar
HueyLD
Posts: 5164
Joined: Mon Jan 14, 2008 10:30 am

Post by HueyLD » Mon Sep 07, 2009 4:48 pm

................
Last edited by HueyLD on Sun Feb 08, 2015 7:24 am, edited 2 times in total.

Levett
Posts: 4177
Joined: Fri Feb 23, 2007 2:10 pm
Location: upper Midwest

Post by Levett » Mon Sep 07, 2009 5:00 pm

Well said, Huey:

That sounds good until a retiree/pre-retiree starts to think about his ability to timely identify the secular bullish vs. secular sideways trends. Why not just have both (e.g., 50/50 TIPS/nominal) so that both “inflation” and “no inflation” risks are covered?

Such trends are seen in retrospect.

This retiree plays both ends against the middle. Bob U.
There are some things that count that can't be counted, and some things that can be counted that don't count.

User avatar
speedbump101
Posts: 997
Joined: Thu Oct 18, 2007 10:54 pm
Location: Alberta Canada

Post by speedbump101 » Wed Sep 09, 2009 10:21 am

I have read the book twice, and a couple of things jumped out at me.

Firstly the active like management he discusses in chapters 24 to 28 are contrary to my philosophy, and having lived in that world at one time, I’m very unlikely to employ anything from those chapters in our portfolio again. Granted his research here is interesting to read, however for me it detracts from the overall quality of the book, when viewed wearing Boglehead glasses.

Secondly, while I really appreciate the clear cut manner he has defined when one has to export portfolio risk, ie. grey and red zones, I feel that some of the annuity products he spends considerable time discussing just aren’t practical in the post 2008 world… He mentions this, however in doing a little further research about annuities with GMWBL (guaranteed minimum withdrawal benefits for life) etc. these types of products see to be going the way of the Dodo bird. So while I totally agree with his concept of exporting risk, I think some of his tools might be outdated... If I were looking for an annuity tool right now I'd be inclined towards the simpler ones, and would definitely consult someone with considerable expertise here. From my perspective these products are mostly one time (for life) decisions, and if you don't fully understand them going in I think you could get badly burned. Fortunately at the current time they do not factor into our retirement plans.

So, from my perspective it’s still an excellent read, with (as always) a few head scratchers for anyone who has spent a few decades investing. I still can’t believe he gave away the first 5000 downloads for free, and for us in this lucky group, once again ‘thank you Jim’… and a mere $4 for those who missed the ‘Boxing Day sale.’

If you are approaching retirement, or are currently retired, read this book. It will focus your mind far better than a cup of coffee at 6AM, or 4AM for some “farmers” here :-)

SB…
"Man is not a rational animal, he is a rationalizing animal" -Robert A. Heinlein

Post Reply