Trinity Study Authors update their results

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Locked
User avatar
wade
Posts: 539
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Trinity Study Authors update their results

Post by wade » Thu Mar 31, 2011 6:41 pm

I had a blog entry in which I updated the results of the Trinity study through 2009

http://wpfau.blogspot.com/2010/10/trini ... rawal.html

But now the original Trinity study authors have a new article out which does the same thing:

http://www.fpanet.org/journal/CurrentIs ... cessRates/

I haven't had a chance to read the whole article yet, but this line from their executive summary is a bit surprising:

"We conclude that if 75 percent success is where to draw the line on portfolio success rates, a client can plan to withdraw a fixed amount of 7 percent of the initial value of portfolios composed of at least 50 percent large-company common stocks."

Well, the actual result is not suprising, but it is surprising that they would emphasize it. Taking a 25% chance for failure does sound a bit risky for my tastes.

And I wonder why they still insist on using corporate bonds. Not many others do that. Why would someone want a retirement portfolio consisting only of stocks and corporate bonds?

It should be an interesting read...

User avatar
magellan
Posts: 3469
Joined: Fri Mar 09, 2007 4:12 pm

Re: Trinity Study Authors update their results

Post by magellan » Thu Mar 31, 2011 6:50 pm

wade wrote:And I wonder why they still insist on using corporate bonds. Not many others do that. Why would someone want a retirement portfolio consisting only of stocks and corporate bonds?
Some of us still think corporate bonds are their own unique asset class :)

I know many around here think treasuries or TIPS are a better bet. Take equity risk with stocks and all that. But my gut tells me that corporate bonds subject investors to unique risks and that will be rewarded. I don't subscribe to the idea that corporate bonds are just "equity risk" plus treasuries.

IMO, corporates and treasuries are entirely different beasts. Still, I'm not a fan of just having stocks and corporate bonds in the portfolio, but I do beleive corp bonds are an important asset class to have in the mix.

Jim

PS - Thanks for the update on Trinity...

User avatar
Taylor Larimore
Advisory Board
Posts: 27507
Joined: Tue Feb 27, 2007 8:09 pm
Location: Miami FL

New Trinity Study

Post by Taylor Larimore » Thu Mar 31, 2011 7:46 pm

Hi Wade:

Thank you for bringing the updated Trinity Study to our attention. I used the earlier Trinity Study for our own retirement planning.

In my opinion, Table 2 in this newer study (Table 3 in the earlier study) provides a very good and easy understanding of safe, inflation-adjusted withdrawal rates.
"Simplicity is the master key to financial success." -- Jack Bogle

User avatar
Peter Foley
Posts: 4582
Joined: Fri Nov 23, 2007 10:34 am
Location: Lake Wobegon

Post by Peter Foley » Thu Mar 31, 2011 8:21 pm

Great article. The "Literature Review" at the end is worth reading as well. It summarizes many of the studies that examined ways to increase withdrawal rates safely.

User avatar
Random Musings
Posts: 5221
Joined: Thu Feb 22, 2007 4:24 pm
Location: Pennsylvania

Post by Random Musings » Thu Mar 31, 2011 8:25 pm

Still looks like that 4% range with at least 25% in equities is still a reasonable approach. 3% withdrawal rate would be very conservative, according to their study.

RM

User avatar
wade
Posts: 539
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Post by wade » Thu Mar 31, 2011 9:40 pm

I read through the article now and found it kind of disappointing. It is simply an update of their 1998 article.

Nisiprius should be happy, because they make more explicit now his point about the need for adaptive withdrawals in response to market conditions by stating more explicitly, "It is not a contractual obligation; it is not a guaranteed annuity rate."

But their system for adaptive withdrawals seems a bit primitive compared to a lot of the research that has come since 1998. They are essentially saying to use their tables to do a reset: to start over with a new withdrawal rate whenever you experience unusually good or bad returns for your portfolio.

They also write, "Because Tables 1 and 2 reflect the bear market of 2008–2009, they are especially useful in that circumstance." However, researchers now understand clearly that portfolio returns at the beginning of retirement are of much greater importance than returns at the very end of retirement. The bad returns of 2008 come at the tail end of retirements and have very little effect.

Anyone looking at the results for 30-year retirements from the Trinity study must clearly understand that it only incorporates retirements beginning up to 1980. Once you think about it, it is obvious: they cannot calculate the results over 30 years for anyone retiring after 1980. I am quite concerned that retirees since the mid 1990s are going to find that the 4% rule, as it is traditionally defined, will not be sustainable.

And as this is not their initial study anymore, there is really no excuse for them to continue to ignore administrative fees. Administrative fees have a huge impact on the results. When I get a chance, I will re-create the table of their inflation-adjusted probabilities after incorporating fees in order to show this.

Just one final note, in their new study they switched to using monthly data rather than annual data like before. This explains small differences between their new Table 2 and the table from my blog entry.
Last edited by wade on Wed Apr 27, 2011 12:45 am, edited 1 time in total.

User avatar
1210sda
Posts: 1408
Joined: Wed Feb 28, 2007 8:31 am

Post by 1210sda » Thu Mar 31, 2011 9:53 pm

Wade, I too was a little disappointed, but thanks for putting it so much more eloquently than I could have.

1210

User avatar
Lbill
Posts: 4997
Joined: Thu Mar 13, 2008 11:25 pm
Location: Somewhere between Up and Down

Post by Lbill » Fri Apr 01, 2011 4:26 am

Nice review Wade. I find the data presented in the study very informative in a few ways:

First, we see that the most critical factors in determining success rates for withdrawals adjusted for cost of living are (1) initial withdrawal rate, (2) years of payout, and (3) the guesstimation of portfolio failure one is willing to accept.

If the withdrawal rate is low enough, 3% - 4%, then the likelihood of portfolio failure is fairly insensitive to asset allocation between equities and bonds.

If the payout period is short enough (20 years or less), portfolio failure is relatively insensitive to asset allocation, up to inflation-adjusted withdrawal rates of 5% or so.

The main effect of asset allocation (percent equities) is to improve the odds of guesstimated portfolio survival at higher withdrawal rates (5% or more) and for longer payout periods (20 years or greater).

With respect to accommodating longer payout periods it should be observed, however, that the results of the study assume that a given equity allocation is maintained throughout the payout period; e.g. 50%. This would mean that a retiree who begins retirement at age 65 with 50% of his nestegg in stocks, in order to afford a 5% real withdrawal rate for a 30-year life expectancy, would need to plan to maintain that same 50% stock allocation at, say, age 80, 85, or even 90 in order to realize the outcome expected on the basis of the reported research results of the study. Certainly contrary to the "age in bonds" mantra and risky by most anyone's standards.

Also, it should be observed that, by holding a large portion of one's retirement portfolio in risky stocks in order to afford a higher withdrawal rate, one has entered the realm of what Arnott refers to as the dangerous game of "probability chicken." Yes, the results of the study suggest that you could have gotten away with a 5% or greater real withdrawal rate during 75% or 80% of historical 30-year periods. So the odds seem to be in your favor when playing this game. But the misses, if and when they occur, can be devastating. For example, large equity losses in the early years of retirement drawdown are sure to instigate a "miss," and you have exposed yourself to this risk.

Finally, Table 4 illustrates the difficulty of walking the fine line between running out of money too soon, and leaving the casino with a large pile of chips left on the table. In order to realize the fruits of your expected success with a high portfolio equity allocation, you have to push your withdrawal rate to as high a level as you dare - playing the game of probability chicken. If you don't do this, and "chicken out" by not pushing the edge of the withdrawal rate envelope then, if there is a beneficiary of your risky retirement investment strategy, it isn't you. You played the game and someone else gets to collect the winnings.
Last edited by Lbill on Fri Apr 01, 2011 6:40 am, edited 1 time in total.
"Life can only be understood backward; but it must be lived forward." ~ Søren Kierkegaard | | "You can't connect the dots looking forward; but only by looking backwards." ~ Steve Jobs

User avatar
jidina80
Posts: 729
Joined: Wed Feb 24, 2010 5:05 pm
Location: Fiji

Re: Trinity Study Authors update their results

Post by jidina80 » Fri Apr 01, 2011 4:33 am

wade wrote:...Taking a 25% chance for failure does sound a bit risky for my tastes.
That 25% is the chance that the portfolio won't last 30 years. It is not the probabilty that it fails during the retiree's life. It is more important to estimate the probability that it will fail before one dies.

According to Vanguard, there is only an 18% chance that one person in a 65 year-old man-woman couple will live to age 95 (other estimates put it at 23%). The math gets a little complex, but probability of a problem with the above 25% failure porfolio is really a lot smaller.

Just.

jimkinny
Posts: 1272
Joined: Sun Mar 14, 2010 1:51 pm

Post by jimkinny » Fri Apr 01, 2011 6:23 am

Thanks for the link

jim

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

Post by bob90245 » Fri Apr 01, 2011 6:54 am

Lbill wrote:With respect to accommodating longer payout periods it should be observed, however, that the results of the study assume that a given equity allocation is maintained throughout the payout period; e.g. 50%. This would mean that a retiree who begins retirement at age 65 with 50% of his nestegg in stocks, in order to afford a 5% real withdrawal rate for a 30-year life expectancy, would need to plan to maintain that same 50% stock allocation at, say, age 80, 85, or even 90 in order to realize the outcome expected on the basis of the reported research results of the study. Certainly contrary to the "age in bonds" mantra and risky by most anyone's standards.
Haven't read the article yet. But it seems to me that based on existing SWR historical studies, 5% by itself is more risky, regardless of constant 50% stock allocation, let alone "age in bonds" glidepath. And as already been revealed from existing historical studies by Bengen, generally "age in bonds" glidepath will require lower initial withdrawal rate than the constant fixed stock allocation.
Ignore the market noise. Keep to your rebalancing schedule whether that is semi-annual, annual or trigger bands.

User avatar
wade
Posts: 539
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Post by wade » Fri Apr 01, 2011 7:26 am

bob90245 wrote:But it seems to me that based on existing SWR historical studies, 5% by itself is more risky, regardless of constant 50% stock allocation, let alone "age in bonds" glidepath. And as already been revealed from existing historical studies by Bengen, generally "age in bonds" glidepath will require lower initial withdrawal rate than the constant fixed stock allocation.
Yes, an aspect of the U.S. historical data (is it an anomaly, or will it endure?) is that higher stock allocations have tended to support higher sustainable withdrawal rates. Historically, this would have worked to the detriment of "age in bonds" style strategies.

I just wanted to mention again one point, which I don't think I said clearly earlier. With the experience from my blog, I already know that a natural reaction to the updated study is going to be:

"Wow, look at that, even after the lost decade of the 2000s, the 4% rule endures. The probabilities of success even increased! It is a really robust finding and I don't know why anyone is worried about it!"

But the problem with this interpretation is that the lost decade only came at the end of retirement periods, not at the beginning. This is what I was trying to get at when I said the Trinity study only considers 30-year retirements starting up to 1980. Were the lost decade to cause the 4% rule to fail, it will be for new retirees from the last decade, not for the people who retired in the late 1970s. Given how the Trinity study works, we will not know if there is going to be an increased failure rate for another 20-30 years.

Also, regarding the fees. I am still using annual data like their original study, rather than monthly data like their updated study. But let me show their original Table 3 / revised Table 2 (it is the table with inflation adjustments) after incorporating some administrative fees.

If retirees pay a 1% account fee at the end of the year, the table looks like this:

Code: Select all

1% Administrative Fee
               3%   4%   5%   6%   7%   8%   9%  10%  11%  12%
100% Stocks
15 Years       100  100  100   87   81   73   69   57   49   41
20 Years       100  100   88   75   66   60   46   40   31   22
25 Years       100   93   77   65   55   47   37   30   20    8
30 Years       100   89   69   56   45   35   29   20    9    5
75% Stocks
15 Years       100  100  100   93   81   71   60   51   46   27
20 Years       100  100   88   75   63   51   42   32   17    9
25 Years       100   93   77   62   48   38   28   12    7    2
30 Years       100   89   65   49   36   20    7    2    0    0
50% Stocks
15 Years       100  100  100   90   79   69   50   39   34   14
20 Years       100  100   86   71   58   37   28   12    6    2
25 Years       100   92   70   52   32   18   10    7    2    2
30 Years       100   84   53   31   13    2    0    0    0    0
25% Stocks
15 Years       100  100  100   86   66   49   34   33   19    9
20 Years       100   97   75   49   31   25   17    8    5    2
25 Years       100   80   45   28   13   12    8    2    2    0
30 Years       100   51   24   11    4    0    0    0    0    0
0% Stocks
15 Years       100  100   94   67   37   34   31   24   14    6
20 Years       100   80   40   29   28   23   11    6    2    2
25 Years        88   38   28   18   13    8    5    2    2    0
30 Years        42   25   13    4    0    0    0    0    0    0
And with 2% account fees:

Code: Select all

2% Administrative Fee
               3%   4%   5%   6%   7%   8%   9%  10%  11%  12%
100% Stocks
15 Years       100  100   94   83   76   70   63   51   46   34
20 Years       100   94   78   71   62   52   43   34   22   15
25 Years       100   88   72   60   48   38   32   23   10    5
30 Years       100   78   58   53   38   31   16    9    5    2
75% Stocks
15 Years       100  100   97   86   74   71   54   49   37   23
20 Years       100   95   80   69   57   45   34   18   11    3
25 Years       100   87   68   55   40   30   15    7    2    0
30 Years       100   76   58   42   24    7    0    0    0    0
50% Stocks
15 Years       100  100   99   84   71   60   46   39   20    6
20 Years       100   95   80   63   43   32   17    6    3    2
25 Years       100   83   57   40   20   10    7    2    2    0
30 Years        96   65   42   16    2    0    0    0    0    0
25% Stocks
15 Years       100  100   99   76   59   43   34   20   10    4
20 Years       100   89   60   42   29   17    9    6    2    2
25 Years        97   57   33   18   12    8    2    2    0    0
30 Years        76   31   15    4    0    0    0    0    0    0
0% Stocks
15 Years       100  100   81   51   37   33   27   16    9    1
20 Years       100   65   35   29   25   11    8    2    2    0
25 Years        60   32   22   13    8    5    2    2    0    0
30 Years        36   18    4    2    0    0    0    0    0    0
This shouldn't be a big problem for Bogleheads, but it could be for many retirees.

The 50-50 portfolio over 30 years with 4% withdrawals had a 96% success rate without fees, 84% success rate with 1% fees, and 65% success rate with 2% fees. This, in and of itself, is a reason to be cautious about using 4%.

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

Post by bob90245 » Fri Apr 01, 2011 10:24 am

wade wrote:... the lost decade only came at the end of retirement periods, not at the beginning. This is what I was trying to get at when I said the Trinity study only considers 30-year retirements starting up to 1980. Were the lost decade to cause the 4% rule to fail, it will be for new retirees from the last decade, not for the people who retired in the late 1970s.
Wade, your comment above is taking a far too literal takeaway. In the strict, academic sense, you are correct.

However, in actual practice, I doubt many retirees religously follow the exact inflation-adjustment withdrawal as prescribed by the Trinity study each and every year through their retirement. If I am off-base on this, please let me know.

My sense is that retirees will follow a more flexible, performance-based withdrawal scheme. In other words, if they find the portfolio balance is being drawn down too rapidly, they will withdraw less. Conversely, if the portfolio balance grows far above expectations, they will spend more freely.

So what are recent retirees who experienced the lost decade of the 2000's doing? Perhaps withdrawing a little less?
Ignore the market noise. Keep to your rebalancing schedule whether that is semi-annual, annual or trigger bands.

Snowjob
Posts: 1511
Joined: Sun Jun 28, 2009 10:53 pm

Post by Snowjob » Fri Apr 01, 2011 10:26 am

wade wrote:
bob90245 wrote:I just wanted to mention again one point, which I don't think I said clearly earlier. With the experience from my blog, I already know that a natural reaction to the updated study is going to be:

"Wow, look at that, even after the lost decade of the 2000s, the 4% rule endures. The probabilities of success even increased! It is a really robust finding and I don't know why anyone is worried about it!"

But the problem with this interpretation is that the lost decade only came at the end of retirement periods, not at the beginning. This is what I was trying to get at when I said the Trinity study only considers 30-year retirements starting up to 1980. Were the lost decade to cause the 4% rule to fail, it will be for new retirees from the last decade, not for the people who retired in the late 1970s. Given how the Trinity study works, we will not know if there is going to be an increased failure rate for another 20-30 years.
This is perhaps one of the most important points and clearly the most over looked that the Trinity study makes. Thanks for pointing it out again, I saw it in your last post but it was a bit burried :D .

At the end of the day, asset allocation is great, but people need to remember VALUATIONS MATTER ! Withdrawing 4% from a 60/40 portfolio when PE's are in the 50s (tech bubble) isnt going to turn out as well. Its not good enough to just pick a conservative AA, w.draw 4% and expect Trinity type success rates.

User avatar
White Coat Investor
Posts: 13524
Joined: Fri Mar 02, 2007 9:11 pm
Location: Greatest Snow On Earth

Post by White Coat Investor » Fri Apr 01, 2011 10:41 am

Snowjob wrote:
wade wrote:
bob90245 wrote:I just wanted to mention again one point, which I don't think I said clearly earlier. With the experience from my blog, I already know that a natural reaction to the updated study is going to be:

"Wow, look at that, even after the lost decade of the 2000s, the 4% rule endures. The probabilities of success even increased! It is a really robust finding and I don't know why anyone is worried about it!"

But the problem with this interpretation is that the lost decade only came at the end of retirement periods, not at the beginning. This is what I was trying to get at when I said the Trinity study only considers 30-year retirements starting up to 1980. Were the lost decade to cause the 4% rule to fail, it will be for new retirees from the last decade, not for the people who retired in the late 1970s. Given how the Trinity study works, we will not know if there is going to be an increased failure rate for another 20-30 years.
This is perhaps one of the most important points and clearly the most over looked that the Trinity study makes. Thanks for pointing it out again, I saw it in your last post but it was a bit burried :D .

At the end of the day, asset allocation is great, but people need to remember VALUATIONS MATTER ! Withdrawing 4% from a 60/40 portfolio when PE's are in the 50s (tech bubble) isnt going to turn out as well. Its not good enough to just pick a conservative AA, w.draw 4% and expect Trinity type success rates.
Yes, but what we need to know is how much of a difference it makes. Are you still okay at 3%? Does it just lower the likelihood of success with 4%withdrawal to 92% instead of 96%? It seems to me that some on this site advocating a 2% withdrawal rate are just being hyper-conservative. But perhaps the numbers will show me to be wrong.
1) Invest you must 2) Time is your friend 3) Impulse is your enemy | 4) Basic arithmetic works 5) Stick to simplicity 6) Stay the course

User avatar
Dick Purcell
Posts: 520
Joined: Tue Oct 26, 2010 1:58 am

Post by Dick Purcell » Fri Apr 01, 2011 10:43 am

Wade, have you or others tried the same rolling-history testing going through the years backwards? EG, 2010, 2009, 2008 . .

One could certainly scream “the past did not go in that direction!” Still, testing backwards could be informative. It’s using the same actual annual rates. Presumably, to the same extent as going forward, it captures your point of “reversion” – that is, high sets of years tend to be followed (and preceded) by low sets of years and vice versa. AND – it enables you to have your test include sequences with the most-important “early years” being those low-return-rate years of the past decade.

Dick Purcell

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

Post by bob90245 » Fri Apr 01, 2011 10:45 am

Snowjob wrote:At the end of the day, asset allocation is great, but people need to remember VALUATIONS MATTER ! Withdrawing 4% from a 60/40 portfolio when PE's are in the 50s (tech bubble) isnt going to turn out as well. Its not good enough to just pick a conservative AA, w.draw 4% and expect Trinity type success rates.
Which begs the question. The outsize returns in the late 1990's gave a huge boost to many retirement portfolios. Much more so had the tech bubble not occurred. So what should a person ready to retire do given they've just reached their "Number" in 2000?
Ignore the market noise. Keep to your rebalancing schedule whether that is semi-annual, annual or trigger bands.

FinanceGeek
Posts: 835
Joined: Sun Jul 01, 2007 5:27 pm

Post by FinanceGeek » Fri Apr 01, 2011 11:07 am

It surprises me that they haven't extended the study to periods beyond 30 years. They have 14 more years of data versus when the Trinity study was originally published...

Snowjob
Posts: 1511
Joined: Sun Jun 28, 2009 10:53 pm

Post by Snowjob » Fri Apr 01, 2011 11:28 am

EmergDoc wrote:
Snowjob wrote:Yes, but what we need to know is how much of a difference it makes. Are you still okay at 3%? Does it just lower the likelihood of success with 4%withdrawal to 92% instead of 96%? It seems to me that some on this site advocating a 2% withdrawal rate are just being hyper-conservative. But perhaps the numbers will show me to be wrong.
bob90245 wrote:Which begs the question. The outsize returns in the late 1990's gave a huge boost to many retirement portfolios. Much more so had the tech bubble not occurred. So what should a person ready to retire do given they've just reached their "Number" in 2000?
I think the answer is to realize you haven't actually made your number. If we take a historical average PE of 15 and then apply that to the hypothetical 60% stock allocation during the peak, we'll find that should reversion to the mean occur well likely lose 40% of the value of the portfolio. This should signal to the investor in my mind 1, change your allocation to reflect the obscene valuations or 2, pretend that your only 2/3rds of the way to "your number" and keep on plugging away. I'm an option 1 guy myself.

*Im really screwing up the quote html today or something!

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

Post by bob90245 » Fri Apr 01, 2011 11:31 am

By the way, since we are talking about the hypothetical Year 2000 retiree, I can offer you my research on the topic:

The Year 2000 Retiree
Introduction

With total returns for the S&P 500 flat for the decade of the 2000’s, many wonder if the hypothetical retiree who began retirement at the start of the decade is faring badly. The logic makes sense. Historical data has shown that poor returns early in retirement have lead to lower average withdrawal rates.

In this article, I will display charts showing retirement portfolios of past years that started poorly. Then I will compare those portfolio balances with the portfolio balance for the Year 2000 retiree.
They say a picture is worth a thousand words. :D

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

Snowjob
Posts: 1511
Joined: Sun Jun 28, 2009 10:53 pm

Post by Snowjob » Fri Apr 01, 2011 11:45 am

bob90245 wrote:By the way, since we are talking about the hypothetical Year 2000 retiree, I can offer you my research on the topic:

The Year 2000 Retiree
Introduction

With total returns for the S&P 500 flat for the decade of the 2000’s, many wonder if the hypothetical retiree who began retirement at the start of the decade is faring badly. The logic makes sense. Historical data has shown that poor returns early in retirement have lead to lower average withdrawal rates.

In this article, I will display charts showing retirement portfolios of past years that started poorly. Then I will compare those portfolio balances with the portfolio balance for the Year 2000 retiree.
They say a picture is worth a thousand words. :D

Image
So what we can tell is that in the first 10 years for each of these really bad times to retire you've gone through about 40% (or more) of your portfolio. Best case, down 40%, anything less than a 3% after tax real return annualized and you run out of money for the 30 year window. On the worst case year, down about 500K. you will need a 5% real return to run out right at year 30.

Good thing those early portfolios had the great bull market of the 80s and 90s to keep the up! Who knows what the future holds in store for our hypothetical y2k retiree
Last edited by Snowjob on Fri Apr 01, 2011 11:48 am, edited 1 time in total.

User avatar
grayfox
Posts: 4934
Joined: Sat Sep 15, 2007 4:30 am

Re: Trinity Study Authors update their results

Post by grayfox » Fri Apr 01, 2011 11:47 am

wade wrote:
But now the original Trinity study authors have a new article out which does the same thing:

http://www.fpanet.org/journal/CurrentIs ... cessRates/
So if I am interpreting Table 2 correctly, if you had withdrawn inflation-adjusted 4% from 75% large-cap stocks / 25% corporate bond it was successful 100% of the 30-year periods from 1926 to 1980.

If you went to 100/0 it had 98% success rate.
If you went to 50/50 it had 96% success rate.

So 75/25 is the sweet spot for 4% inflation-adjusted withdrawal rate over 30-year period. And this has not changed since 1998.

ResNullius
Posts: 2091
Joined: Wed Oct 24, 2007 3:22 pm

Post by ResNullius » Fri Apr 01, 2011 11:49 am

It is indeed interesting to see so much of their focus on a 75% success rate as being a reasonable starting point for financial planning during retirement. Given all the uncertainties in planning for the future, particularly out in time for 30 years or more, I would tend to think that 100% is a minimum starting point, unless of course a person simply couldn't live off that amount.

conundrum
Posts: 825
Joined: Sat May 09, 2009 7:00 pm

Post by conundrum » Fri Apr 01, 2011 12:00 pm

Interesting paper and discussion. We plan to withdraw a fixed, not inflation adjusted, % from our portfolio so I was interested to see the data in table one. With our asset allocation at 40% equity and 60% fixed income the closest allocation (50/50) had a 98% success rate at 30 years with a 6% annualized withdrawal rate. Our plan calls for a 3% fixed withdrawal rate and we are not planning to change our withdrawal rate based on this study but interesting data all the same.

Drum :lol:

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

Post by bob90245 » Fri Apr 01, 2011 12:06 pm

Snowjob wrote:So what we can tell is that in the first 10 years for each of these really bad times to retire you've gone through about 40% (or more) of your portfolio. Best case, down 40%, anything less than a 3% after tax real return annualized and you run out of money for the 30 year window. On the worst case year, down about 500K. you will need a 5% real return to run out right at year 30.

Good thing those early portfolios had the great bull market of the 80s and 90s to keep the up! Who knows what the future holds in store for our hypothetical y2k retiree
That's about right. Although, 5% real return over a 30-year period for a 50/50 portfolio was not so unusual.

Image
Source: http://www.bobsfinancialwebsite.com/dow ... Allocation

As an aside, I see the 1965-1994 30-year period had an annual real return below 4% for a 50/50 portfolio.
Ignore the market noise. Keep to your rebalancing schedule whether that is semi-annual, annual or trigger bands.

User avatar
wade
Posts: 539
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Post by wade » Fri Apr 01, 2011 12:13 pm

bob90245 wrote:Wade, your comment above is taking a far too literal takeaway. In the strict, academic sense, you are correct.

However, in actual practice, I doubt many retirees religously follow the exact inflation-adjustment withdrawal as prescribed by the Trinity study each and every year through their retirement. If I am off-base on this, please let me know.

My sense is that retirees will follow a more flexible, performance-based withdrawal scheme. In other words, if they find the portfolio balance is being drawn down too rapidly, they will withdraw less. Conversely, if the portfolio balance grows far above expectations, they will spend more freely.

So what are recent retirees who experienced the lost decade of the 2000's doing? Perhaps withdrawing a little less?
I do understand that actual retirees need to be flexible. Also, I understand that the idea of planning to hit zero wealth precisely at the end of 30 years is not realistic, because once you are 29 years into retirement, you probably want to be planning further ahead. But the point of these types of studies is to get some guidance about what will be feasible so that you can get yourself started on a sustainable path and not have to make drastic changes in the future. At least, that is my starting point. Hypothetically, if we could say that the 4% rule will not be sustainable, I think this is information that new retirees would like to know so that they can plan in anticipation that their future spending goals may not be feasible. So while I understand that "constant inflation-adjusted withdrawals" is not actually realistic, I still find it to be a quite useful starting point for analysis. One needs to be flexible, but the question is: How flexible? What kind of adjustments should people make? If the 4% rule is going to be sustainable, then it is not such a big deal. But if the 4% rule will not be sustainable, then this question is really important.


EmergDoc wrote:Yes, but what we need to know is how much of a difference it makes. Are you still okay at 3%? Does it just lower the likelihood of success with 4%withdrawal to 92% instead of 96%? It seems to me that some on this site advocating a 2% withdrawal rate are just being hyper-conservative. But perhaps the numbers will show me to be wrong.
Here is my earnest attempt to answer this question:
Image
source: http://ideas.repec.org/p/pra/mprapa/27487.html

This study attempts to quantify whether a 4 percent withdrawal rate can still be considered as safe for U.S. retirees in recent years when earnings valuations have been at historical highs and the dividend yield has been at historical lows. We find that the traditional 4 percent withdrawal rule is likely to fail for recent retirees. The maximum sustainable withdrawal rate (MWR) for retirees may continue declining even after the peak in earnings valuations in 2000. Our lowest point estimate for an MWR with a 60/40 allocation between stocks and bonds is 1.46 percent for new retirees in 2008. We also discuss confidence intervals for these predictions. The regression framework with variables to predict long-term stock returns, bond returns, and inflation (the components driving the retiree's remaining portfolio balance) produces estimates that fit the historical data quite well, and we use backtesting for a further robustness check. Nevertheless, there are important qualifications for these predictions. In particular, they depend on out-of-sample estimates as the circumstances of the past 15 years have not been witnessed before, and there is always potential for structural changes which could leave recent retirees in better shape than suggested by the model. Looking forward, this methodology can guide new retirees toward a reasonable range for their MWR so that the 4 percent rule need not be blindly followed.

Dick Purcell wrote:Wade, have you or others tried the same rolling-history testing going through the years backwards? EG, 2010, 2009, 2008 . .
That's a good idea, Dick. So, here is the Trinity study inflation-adjusted withdrawals success rate table, but with historical data running in reverse rather than in the proper direction:

Code: Select all

Historical Data Flipped Upside Down

               3%   4%   5%   6%   7%   8%   9%  10%  11%  12%
100% Stocks
15 Years       100  100   99   91   84   80   69   53   51   43
20 Years       100  100   91   83   75   69   54   46   37   26
25 Years       100   98   85   78   75   62   47   35   28   25
30 Years       100   96   82   75   65   51   38   33   25   18
75% Stocks
15 Years       100  100   99   94   89   77   66   54   43   31
20 Years       100  100   94   85   75   66   51   35   28   15
25 Years       100   97   87   78   72   53   38   25   12    5
30 Years       100   93   82   73   60   47   29   13    5    4
50% Stocks
15 Years       100  100  100   93   87   74   60   50   33   16
20 Years       100   98   92   78   66   57   32   15    9    5
25 Years       100   93   83   77   57   35   12    7    5    2
30 Years       100   89   78   65   47   16    7    5    0    0
25% Stocks
15 Years       100  100   99   94   74   63   41   30   14   10
20 Years       100   97   89   68   48   31   20    9    3    0
25 Years        98   90   73   45   30   22    7    2    0    0
30 Years        96   80   55   36   24    7    2    0    0    0
0% Stocks
15 Years       100  100   97   79   53   40   34   23    9    3
20 Years       100   95   55   40   37   26   11    0    0    0
25 Years        97   57   40   35   27   10    0    0    0    0
30 Years        84   45   38   31   13    0    0    0    0    0
The 96% success rate for the 50-50 30-year 4% case changed to 89%.
bob90245 wrote:Which begs the question. The outsize returns in the late 1990's gave a huge boost to many retirement portfolios. Much more so had the tech bubble not occurred. So what should a person ready to retire do given they've just reached their "Number" in 2000?
Actually, the reverse of this question led me toward writing a paper on "safe savings rates".
http://ideas.repec.org/p/pra/mprapa/28796.html

Mr. Larimore mentioned before that he uses the Trinity study for his own retirement planning. I also know he retired in 1982. I was thinking to say something like: well, given that you retired in a year that will have one of the highest sustainable withdrawal rates in history (just under 9% in my figure above) you could hardly have gone wrong with your withdrawals. But before making the comment, I realized it was wrong, because the 1982 retiree entered retirement after years of seeing their portfolio get smashed by negative returns and inflation. The 1982 retiree can enjoy a high withdrawal rate, but from a lower starting wealth accumulation.

Likewise, the 2000 retiree should have enjoyed accumulating one of the highest wealth accumulations in history, which should provide some cushion should they experience lower sustainable withdrawal rates. But I worry if things will really work out that way for 2000 retirees. They may have quit saving as soon as they hit their magic number. As I describe in my paper, with my safe savings rate approach, the 2000 retiree will be okay as long as their sustainable withdrawal rate ends up being above 2.62 percent. I'm worried it may not even be that high, but the safe savings rate would still have helped get them closer to the sustainable path than just using a 4% rate.
FinanceGeek wrote:It surprises me that they haven't extended the study to periods beyond 30 years. They have 14 more years of data versus when the Trinity study was originally published...
Here you go, the same table, but with 35 year and 40 year retirements also included:

Code: Select all

With 35 Years and 40 Years too

               3%   4%   5%   6%   7%   8%   9%  10%  11%  12%
100% Stocks
15 Years       100  100  100   93   83   76   70   64   54   47
20 Years       100  100   91   80   71   63   54   46   40   31
25 Years       100  100   87   73   63   52   42   37   28   20
30 Years       100   96   78   62   55   42   33   29   20   11
35 Years       100   92   72   56   50   36   30   24   12    6
40 Years       100   91   76   58   51   38   31   24   11    7
75% Stocks
15 Years       100  100  100   96   86   74   71   57   49   41
20 Years       100  100   92   80   69   60   49   40   29   15
25 Years       100  100   85   68   57   43   35   27   12    7
30 Years       100   98   76   62   45   36   20    5    2    0
35 Years       100   92   68   52   38   28   10    2    0    0
40 Years       100   91   67   49   38   18    7    0    0    0
50% Stocks
15 Years       100  100  100   94   83   71   61   46   39   23
20 Years       100  100   92   80   63   46   35   22    9    6
25 Years       100  100   82   62   45   27   15    8    7    2
30 Years       100   96   69   49   20   11    2    0    0    0
35 Years       100   88   54   34   12    0    0    0    0    0
40 Years       100   82   51   20    9    0    0    0    0    0
25% Stocks
15 Years       100  100  100   91   76   60   46   34   26   14
20 Years       100  100   86   58   45   31   23   15    8    2
25 Years       100   95   58   35   23   13   10    8    2    2
30 Years       100   75   33   22    7    2    0    0    0    0
35 Years       100   50   18   10    2    0    0    0    0    0
40 Years       100   33   18    4    0    0    0    0    0    0
0% Stocks
15 Years       100  100  100   77   51   37   34   29   19   13
20 Years       100   92   58   35   29   26   20    9    5    2
25 Years       100   58   32   23   18   12    8    5    2    2
30 Years        85   35   22   11    2    0    0    0    0    0
35 Years        46   22   12    2    0    0    0    0    0    0
40 Years        29   16    4    0    0    0    0    0    0    0
That 96% for 30 years becomes 88% for 35 years and 82% for 40 years.

Note, in a couple cases the probabilities increase as the time length increases. This is because for 40 years, we can only consider retirements beginning up to 1970. Any retirements between 1971-1980 that failed with 30 years will not show up in the 40 year calculations. This explains it.
Snowjob wrote:I think the answer is to realize you haven't actually made your number. If we take a historical average PE of 15 and then apply that to the hypothetical 60% stock allocation during the peak, we'll find that should reversion to the mean occur well likely lose 40% of the value of the portfolio. This should signal to the investor in my mind 1, change your allocation to reflect the obscene valuations or 2, pretend that your only 2/3rds of the way to "your number" and keep on plugging away. I'm an option 1 guy myself.
This is also the problem I was looking at in my safe savings rate paper. But in that paper, I followed your option 2 instead of your option 1.
http://ideas.repec.org/p/pra/mprapa/28796.html
bob90245 wrote:They say a picture is worth a thousand words.
Yes, but the other aspect of that picture is this:

Image
source: It is not yet in the paper, but will eventually be included in this paper: http://ideas.repec.org/p/pra/mprapa/27107.html

allsop
Posts: 1046
Joined: Sun Jun 15, 2008 7:08 am

Post by allsop » Fri Apr 01, 2011 12:30 pm

EmergDoc wrote: [snip]
Yes, but what we need to know is how much of a difference it makes. Are you still okay at 3%? Does it just lower the likelihood of success with 4%withdrawal to 92% instead of 96%? It seems to me that some on this site advocating a 2% withdrawal rate are just being hyper-conservative. But perhaps the numbers will show me to be wrong.
From my point of view it is prudent to subtract (predictable) expenses like fund and plan fees that often are very high, and not all Bogleheads live in USA where you cheaply can roll over a 401(k) to Vanguard.

allsop
Posts: 1046
Joined: Sun Jun 15, 2008 7:08 am

Post by allsop » Fri Apr 01, 2011 12:35 pm

wade wrote: [snip] And as this is not their initial study anymore, there is really no excuse for them to continue to ignore administrative fees. Administrative fees have a huge impact on the results. When I get a chance, I will re-create the table of their inflation-adjusted probabilities after incorporating fees in order to show this.
Thank you for the tables!

Do you have similar research using a global market cap portfolio? That would be quite useful for those of us not living USA.

User avatar
Mel Lindauer
Moderator
Posts: 28637
Joined: Mon Feb 19, 2007 8:49 pm
Location: Daytona Beach Shores, Florida
Contact:

Post by Mel Lindauer » Fri Apr 01, 2011 12:37 pm

Mr. Larimore mentioned before that he uses the Trinity study for his own retirement planning. I also know he retired in 1982. I was thinking to say something like: well, given that you retired in a year that will have one of the highest sustainable withdrawal rates in history (just under 9% in my figure above) you could hardly have gone wrong with your withdrawals. But before making the comment, I realized it was wrong, because the 1982 retiree entered retirement after years of seeing their portfolio get smashed by negative returns and inflation. The 1982 retiree can enjoy a high withdrawal rate, but from a lower starting wealth accumulation.
As you're pointing out, it's all relative to the starting amount, and that's likely determined by the retiree's previous market returns. So a really good study might account for the previous return sequence prior to retirement, with special emphasis on perhaps the 10 years closest to retirement since that's when the investor would have the greatest amount accumulated.

For example, if we compare two investors who saved and invested approx. the same amount over their working years but ended up with different amounts, based soley on the market returns prior to their retirement. Each needs $40,000.

4% of $1,00,000 = $40,000 per year.
2% of $2,00,000 = $40,000 per year.
Best Regards - Mel | | Semper Fi

User avatar
wade
Posts: 539
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Post by wade » Fri Apr 01, 2011 12:44 pm

Mel Lindauer wrote:
Mr. Larimore mentioned before that he uses the Trinity study for his own retirement planning. I also know he retired in 1982. I was thinking to say something like: well, given that you retired in a year that will have one of the highest sustainable withdrawal rates in history (just under 9% in my figure above) you could hardly have gone wrong with your withdrawals. But before making the comment, I realized it was wrong, because the 1982 retiree entered retirement after years of seeing their portfolio get smashed by negative returns and inflation. The 1982 retiree can enjoy a high withdrawal rate, but from a lower starting wealth accumulation.
As you're pointing out, it's all relative to the starting amount, and that's likely determined by the retiree's previous market returns. So a really good study might account for the previous return sequence prior to retirement, with special emphasis on perhaps the 10 years closest to retirement since that's when the investor would have the greatest amount accumulated.

For example, if we compare two investors who saved and invested approx. the same amount over their working years but ended up with different amounts, based soley on the market returns prior to their retirement.

4% of $1,00,000 = $40,000 per year.
2% of $2,00,000 = $40,000 per year.
Dear Mel,

I think that you are precisely describing the theme of my "safe savings rate" paper that will be published in the May 2011 Journal of Financial Planning.

Your example illustrates my point as well that for some desired retirement expenditure amounts, the safe savings rates are much smoother than traditional retirement planning (choose a withdrawal rate and then try to accumulate enough wealth so that your withdrawal rate matches your desired expenditures).

I linked to it before, but here is my blog entry summarizing it also:
http://wpfau.blogspot.com/2011/02/safe- ... ch-to.html

Please let me know if your idea is different.

User avatar
wade
Posts: 539
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Post by wade » Fri Apr 01, 2011 12:47 pm

allsop wrote:
wade wrote: [snip] And as this is not their initial study anymore, there is really no excuse for them to continue to ignore administrative fees. Administrative fees have a huge impact on the results. When I get a chance, I will re-create the table of their inflation-adjusted probabilities after incorporating fees in order to show this.
Thank you for the tables!

Do you have similar research using a global market cap portfolio? That would be quite useful for those of us not living USA.
Allsop, sorry, I'm not presently able to answer that one. Someday, someday.

I do have this though:

An International Perspective on Safe Withdrawal Rates: The Demise of the 4 Percent Rule?
http://www.fpanet.org/journal/CurrentIs ... awalRates/

Snowjob
Posts: 1511
Joined: Sun Jun 28, 2009 10:53 pm

Post by Snowjob » Fri Apr 01, 2011 12:55 pm

wade wrote:
bob90245 wrote:They say a picture is worth a thousand words.
Yes, but the other aspect of that picture is this:

Image
source: It is not yet in the paper, but will eventually be included in this paper: http://ideas.repec.org/p/pra/mprapa/27107.html
And I'm sure that retiree in 2000, would gladly give up 5% of his portflio's wealth and trade down into the market that the 1940 retiree had. Half the P/E and more than triple the yield -- it all comes back to valuation

ResNullius
Posts: 2091
Joined: Wed Oct 24, 2007 3:22 pm

Post by ResNullius » Fri Apr 01, 2011 3:17 pm

Well, if you're willing to suffer the ups and downs, why would anyone opt for less than 100% in equities. It pays out while you're alive, then pays your heirs very nicely too. Don't get me wrong, I'm 69% in fixed right now, and I plan on staying heavy into fixed now that I'm in the withdrawal years. But, it's still an interesting question based on the data points in the study.

User avatar
Dick Purcell
Posts: 520
Joined: Tue Oct 26, 2010 1:58 am

Post by Dick Purcell » Fri Apr 01, 2011 3:21 pm

Wade, thanks for running that years-in-reverse test. (How in the world can you be responding to us all at that pace?)

With the years running backward, at least for one case there’s astonishing agreement between rolling history and Monte!

Under your running-backward table, you point out that for 50-50 portfolio, 30-year run, 4% WD, the success rate is 89% -- or stated negatively. 11% failure.

WellSir – I extracted a probability distribution from your historical data source, Shiller, and with the same 50-50 portfolio ran 10,000 Monte sims of 30 years with 4% WD, and got 11% failure rate ! !

As we’ve discussed, each method has at least one weakness compared to the other. Monte, the way I used it, has no serial correlation and thus has only random-walk reversion toward the mean. If I understand you correctly, you point out that in the history there is “super” regression – that is, after some years above mean, the next few are more likely to be below mean.

On the other hand, Monte has any number of samples – I find it desirable to run thousands, and as I noted I used 10,000 – while rolling history has vastly fewer, roughly 50 overlappers. My discussion partner Rodc would I think point out that it has only 3 or 4 non-duplicative samples.

Two cautions: (1) For me, and I think for you too, that 11% failure rate is with zero fees. (2) A Fama-French report says that since the middle of the last century, the “super” regression in the history has shrunk. Their report is here: http://www.albany.edu/faculty/faugere/P ... onFama.pdf

Dick Purcell

unclemick
Posts: 1318
Joined: Tue Feb 20, 2007 10:18 am
Location: greater Kansas City

Post by unclemick » Fri Apr 01, 2011 3:50 pm

Yes agile, mobile and hostile.

To clarify - a phrase I use to denote aggressive retirement portfolio defense by varing withdrawal rate. I think it came from Bear Bryant when asked what he wanted to see in an Alabama linebaker.

I use it often.

heh heh heh - 8) Perhaps I should drop mobile as I'm more and more liking Mr. Bogle's Stay The Course as years go by.
Last edited by unclemick on Sat Apr 02, 2011 11:10 am, edited 1 time in total.

User avatar
patriciamgr2
Posts: 788
Joined: Mon Nov 19, 2007 3:06 pm

Wade: thanks so much for informative posts

Post by patriciamgr2 » Fri Apr 01, 2011 4:08 pm

I really appreciate discussions like this & the generous sharing of information and links. It's hard to find intellectually rigorous analyses on what to retirees is the all important question of SWR--this is one of the reasons the boglehead site is so valuable. thx.

User avatar
LadyGeek
Site Admin
Posts: 48576
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Post by LadyGeek » Fri Apr 01, 2011 4:22 pm

Following patriciamgr2's comments, this thread is now in the wiki. Wiki article link: Safe Withdrawal Rates

Is there anything that should be added or referenced from this on-going discussion? The wiki stops at 2008, so perhaps the content should be updated. I don't have the background to do this myself.

Wade - I added your April 1, 2011 blog entry (External links).

Comments / questions / concerns are welcome. Wiki editors can update directly, this is a collaborative effort.
Wiki To some, the glass is half full. To others, the glass is half empty. To an engineer, it's twice the size it needs to be.

User avatar
wade
Posts: 539
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Post by wade » Fri Apr 01, 2011 6:24 pm

Dick Purcell wrote:Wade, thanks for running that years-in-reverse test. (How in the world can you be responding to us all at that pace?)

With the years running backward, at least for one case there’s astonishing agreement between rolling history and Monte!

Under your running-backward table, you point out that for 50-50 portfolio, 30-year run, 4% WD, the success rate is 89% -- or stated negatively. 11% failure.

WellSir – I extracted a probability distribution from your historical data source, Shiller, and with the same 50-50 portfolio ran 10,000 Monte sims of 30 years with 4% WD, and got 11% failure rate ! !

As we’ve discussed, each method has at least one weakness compared to the other. Monte, the way I used it, has no serial correlation and thus has only random-walk reversion toward the mean. If I understand you correctly, you point out that in the history there is “super” regression – that is, after some years above mean, the next few are more likely to be below mean.

On the other hand, Monte has any number of samples – I find it desirable to run thousands, and as I noted I used 10,000 – while rolling history has vastly fewer, roughly 50 overlappers. My discussion partner Rodc would I think point out that it has only 3 or 4 non-duplicative samples.

Two cautions: (1) For me, and I think for you too, that 11% failure rate is with zero fees. (2) A Fama-French report says that since the middle of the last century, the “super” regression in the history has shrunk. Their report is here: http://www.albany.edu/faculty/faugere/P ... onFama.pdf

Dick Purcell
Hi Dick, Thanks for the article link. I need to read that one. About responding so quickly, the program I wrote for the Trinity study allowed me to just make one small change for each of these extra cases and then it spit out the table exactly as seen here. So it was no trouble.

About the super regression, in "Beyond Greed and Fear," Hersh Shefrin points out that it is gambler's fallacy to believe that above-average returns will be followed by below-average returns. However, Campbell and Shiller demonstrate that there is a relationship [though this relationship is somewhat controversial not particularly strong] between valuation measures and subsequent real returns. High returns will tend to push up valuations, which lowers subsequent returns.

For investigations of the retirement period, I would be willing to concede that Monte Carlo is a better approach than historical simulations. I think Monte Carlo makes 100% stocks look less impressive than one observes with the actual sequence of returns in the U.S. historical data.
unclemick wrote:Yes agile, mobile and hostile.
I'm sorry if I came off sounding hostile. I got into a bad mood because one of my major pet peeves is people who publish the same article over and over in different places. The 2011 version of the Trinity Study is so close to the 1998 version that it triggers my pet peeve strongly. I thought they might actually be doing something new in their paper, so I was feeling disappointed. Also, I really like many aspects of the Journal of Financial Planning, but they also had a paper in January that was just a rehash of an earlier study as well.

patriciamgr2 and LadyGeek: Thanks a lot for your nice comments. The new Trinity study should probably be added as a link in the Wiki, as the Wiki entry is mostly centered around it. I personally feel that William Bengen's 1994 paper should be featured more prominently, as I think it was the really innovative study, and the Trinity study followed from his earlier work. About my papers, I think some of them are controversial, but the paper "An International Perspective on Safe Withdrawal Rates: The Demise of the 4 Percent Rule?" doesn't really have any controversy to it, so perhaps it could be a possibility? It is this one:
http://www.fpanet.org/journal/CurrentIs ... awalRates/

Thank you.

User avatar
Taylor Larimore
Advisory Board
Posts: 27507
Joined: Tue Feb 27, 2007 8:09 pm
Location: Miami FL

A noteworthy study in the search for Safe Withdrawal Rates

Post by Taylor Larimore » Fri Apr 01, 2011 7:40 pm

Hi Wade:

I am very impressed with your latest study about Safe Withdrawal Rates linked above. It is scholarly in every way. Bogleheads are honored that you have allowed us to be among the first to read it.

We appreciate the fact you included the wise words of our mentor in your closing line:
It may be tempting to hope that asset returns in the 21st century United States will continue to be as spectacular as in the last century, but Bogle (2009) cautions his readers, “Please, please, please: Don’t count on it.
"Simplicity is the master key to financial success." -- Jack Bogle

User avatar
LadyGeek
Site Admin
Posts: 48576
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Post by LadyGeek » Fri Apr 01, 2011 8:16 pm

Wiki changes incorporated. I added Wade's paper and also took some time for reformatting and link cleanup.

Wiki article link: Safe Withdrawal Rates

Wade - Your paper is linked to the "current" issue of the Journal of Financial Planning. Once next month's issue is released, the URL will become invalid. Is there another web site that has a more permanent link?
Wiki To some, the glass is half full. To others, the glass is half empty. To an engineer, it's twice the size it needs to be.

User avatar
wade
Posts: 539
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Post by wade » Fri Apr 01, 2011 8:43 pm

Taylor: Thank you very much for the nice comments. Though I had known about Mr. Bogle for a long time, as he was always featured prominently in the bookstore of his alma mater that I frequently visited, I just happened to read his book Enough at the same time I was working on that paper, and what he wrote there fit in so well with the theme of that paper. Since then, I've been reading much more of his writings and learning about his positive influence on investing. I'm young enough that I can't hardly imagine a time when low-cost index funds didn't exist (I guess I was born just a little bit after the birth of index funds) and am always shocked to hear that there are people out there that don't actually take advantage of them.

LadyGeek: The new formatting is very nice. I like the box around William Bengen's article. I'm sorry for the confusion, I used to think that link would shortly disappear but then it seemed to persist. Actually, the issue for that is December 2010, and yes, that link might disappear someday. Here is another link for it that should be permanent:

http://ideas.repec.org/p/ngi/dpaper/10-12.html

Thank you so much, Wade

User avatar
LadyGeek
Site Admin
Posts: 48576
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Post by LadyGeek » Fri Apr 01, 2011 8:59 pm

Thank you. The date and link have been corrected.

Wiki article link: Safe Withdrawal Rates
Wiki To some, the glass is half full. To others, the glass is half empty. To an engineer, it's twice the size it needs to be.

edge
Posts: 3332
Joined: Mon Feb 19, 2007 7:44 pm
Location: Great Falls VA

Post by edge » Fri Apr 01, 2011 9:49 pm

It seems ....not to bright... to be 'irritated' that the Trinity Study was updated. To me it is completely natural to update a time-based analysis after more time has passed and data gathered.

And the results are similar - so what? That is as significant as the results being different IMO.

User avatar
wade
Posts: 539
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Post by wade » Fri Apr 01, 2011 10:26 pm

edge wrote:It seems ....not to bright... to be 'irritated' that the Trinity Study was updated. To me it is completely natural to update a time-based analysis after more time has passed and data gathered.

And the results are similar - so what? That is as significant as the results being different IMO.
I'm certainly not irritated that it was updated. They could provide an annual update on their webpage or something, and it would be more than fine with me. Or if they just had a guest column at the journal instead of a full research article, that would be fine too. It is just that, as I've always understood it anyway, the social norm of research publications is that you are not supposed to publish the same thing more than once. It is natural to violate this social norm, as publishing the same thing is a rather easy way to increase one's publication count. Maybe this is something just idiosyncratic to me and most people don't care, I'm not sure. The thing is, I really like the Journal of Financial Planning and its mission, but for most academics it is just a joke. When they publish recycled articles, it really doesn't improve their image in academia. Then again, maybe they have no interest in improving their image in academia. Perhaps an update of the Trinity study is exactly what their readers want. Actually, I think it IS what their readers want. This is not too important of an issue for this forum though...

kyounge1956
Posts: 333
Joined: Mon May 31, 2010 5:39 pm

Post by kyounge1956 » Sat Apr 02, 2011 2:28 am

wade wrote:(snip) About responding so quickly, the program I wrote for the Trinity study allowed me to just make one small change for each of these extra cases and then it spit out the table exactly as seen here. So it was no trouble. (snip)
Does your program also allow you to make one small change and tell us what would be in tables 3 and 4 for the longer retirements? I'm a conservative investor planning on a conservative WR. It's reassuring to me to look at tables 3 & 4 and see that I could take 30 years of 3% withdrawals, adjusted for inflation, from a portfolio with 25% stocks and have a real good chance of having more money at the end of the 30 years than I had at the beginning. Does the same thing hold true for a 40-year retirement?

User avatar
wade
Posts: 539
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Post by wade » Sat Apr 02, 2011 6:56 am

kyounge1956 wrote:
wade wrote:(snip) About responding so quickly, the program I wrote for the Trinity study allowed me to just make one small change for each of these extra cases and then it spit out the table exactly as seen here. So it was no trouble. (snip)
Does your program also allow you to make one small change and tell us what would be in tables 3 and 4 for the longer retirements? I'm a conservative investor planning on a conservative WR. It's reassuring to me to look at tables 3 & 4 and see that I could take 30 years of 3% withdrawals, adjusted for inflation, from a portfolio with 25% stocks and have a real good chance of having more money at the end of the 30 years than I had at the beginning. Does the same thing hold true for a 40-year retirement?
Hi, Yes, I can do this one. My results are a little bit different, again, because I am still basing the calculations on annual data like the original study, but the updated study makes calculations using monthly data.

But you can see that with the 3% withdrawal rate, your wealth should continue to grow after 35 or 40 years for any asset allocation except the no stocks case with inflation adjustments. The 25% stocks case with 3% withdrawals you are looking at does work, even with inflation adjustments. But not with 4% and inflation adjustments.

Please keep in mind that these are the median terminal wealth accumulations, which means you have a 50% chance of accumulating more than this amount, and a 50% chance of accumulating less.

If I may add my own two cents, I worry that recent retirees will be leaning toward the accumulating less side, but here are the Trinity-style numbers without any further editorializing:

Code: Select all

Trinity 2011's Table 3: Median End-of-period Portfolio Values WITHOUT Inflation-Adjustments
With 35 Years and 40 Years too

               3%   4%   5%   6%   7%   8%   9%  10%  11%  12%
100% Stocks
15 Years       3868  3564  3183  2812  2513  2214  1833  1516  1227   944
20 Years       7362  6780  5986  5015  3965  3165  2413  1661   943   337
25 Years      10264  9010  8252  6743  5610  4879  3557  1830   829     0
30 Years      18292 16021 12772 10471  8570  6146  3885  1289     0     0
35 Years      28798 24221 20558 15318 10024  5965  2199     0     0     0
40 Years      41791 35807 30200 24620 17855 11633  2708     0     0     0
75% Stocks
15 Years       3392  3119  2777  2412  2048  1711  1378  1080   802   544
20 Years       5426  4743  4076  3413  2916  2384  1726  1010   291     0
25 Years       8484  7095  5710  4582  3707  2562  1498   492     0     0
30 Years      11183  9822  8270  6334  4642  3311  1390     0     0     0
35 Years      19128 16579 13290 10114  7157  3812   542     0     0     0
40 Years      29593 24251 19664 14805  8837  3532   230     0     0     0
50% Stocks
15 Years       2690  2379  2127  1824  1467  1235   940   645   308    32
20 Years       3760  3100  2609  2197  1728  1212   694   155     0     0
25 Years       4864  4022  3254  2517  1693  1135   324     0     0     0
30 Years       7339  5528  4398  3205  2058   850     0     0     0     0
35 Years      11265  9111  6873  4404  2390   223     0     0     0     0
40 Years      18197 15362 10384  6864  2650     0     0     0     0     0
25% Stocks
15 Years       1726  1491  1267  1030   799   577   348   113     0     0
20 Years       2097  1736  1370   952   628   271     0     0     0     0
25 Years       2716  2116  1564   894   342     0     0     0     0     0
30 Years       3551  2795  2039   896     0     0     0     0     0     0
35 Years       5198  3893  2277  1084     0     0     0     0     0     0
40 Years       7824  5581  3339  1140     0     0     0     0     0     0
0% Stocks
15 Years       1524  1322  1104   886   656   437   220     7     0     0
20 Years       1532  1209   952   566   259     0     0     0     0     0
25 Years       1660  1281   849   226     0     0     0     0     0     0
30 Years       1690  1193   707     0     0     0     0     0     0     0
35 Years       1928  1214   470     0     0     0     0     0     0     0
40 Years       2003  1053    37     0     0     0     0     0     0     0

Code: Select all

Trinity 2011's Table 4: Median End-of-period Portfolio Values with Inflation-Adjustments
With 35 Years and 40 Years too
               3%   4%   5%   6%   7%   8%   9%  10%  11%  12%
100% Stocks
15 Years       3548  3182  2822  2445  2058  1468  1029   540   108     0
20 Years       6750  5891  5123  3328  1917  1243   465     0     0     0
25 Years       8782  7359  5868  4565  2044   516     0     0     0     0
30 Years      13003  9913  7255  4265  1493     0     0     0     0     0
35 Years      19216 13678  8215  3603    27     0     0     0     0     0
40 Years      33936 24154 13814  6274   204     0     0     0     0     0
75% Stocks
15 Years       3061  2660  2242  1922  1406   962   695   342     0     0
20 Years       4788  3724  2990  2127  1181   474     0     0     0     0
25 Years       6178  4503  3354  2013   797     0     0     0     0     0
30 Years       8858  6258  3255  1455     0     0     0     0     0     0
35 Years      12978  7527  3663   574     0     0     0     0     0     0
40 Years      20282 11860  6291     0     0     0     0     0     0     0
50% Stocks
15 Years       2372  2002  1580  1284   956   586   209     0     0     0
20 Years       2988  2405  1849  1204   515     0     0     0     0     0
25 Years       3700  2730  1731   849     0     0     0     0     0     0
30 Years       4984  3096  1530     0     0     0     0     0     0     0
35 Years       7234  3761   463     0     0     0     0     0     0     0
40 Years       9064  3971   220     0     0     0     0     0     0     0
25% Stocks
15 Years       1620  1338  1042   787   527   146     0     0     0     0
20 Years       1763  1298   792   285     0     0     0     0     0     0
25 Years       1958  1132   327     0     0     0     0     0     0     0
30 Years       2218   803     0     0     0     0     0     0     0     0
35 Years       2977    58     0     0     0     0     0     0     0     0
40 Years       3287     0     0     0     0     0     0     0     0     0
0% Stocks
15 Years       1205   892   514   273    21     0     0     0     0     0
20 Years       1032   581   111     0     0     0     0     0     0     0
25 Years        855    54     0     0     0     0     0     0     0     0
30 Years        601     0     0     0     0     0     0     0     0     0
35 Years          0     0     0     0     0     0     0     0     0     0
40 Years          0     0     0     0     0     0     0     0     0     0

riskreward
Posts: 106
Joined: Sat Jan 05, 2008 11:32 am

Post by riskreward » Sat Apr 02, 2011 8:39 am

ResNullius wrote:Well, if you're willing to suffer the ups and downs, why would anyone opt for less than 100% in equities. It pays out while you're alive, then pays your heirs very nicely too. Don't get me wrong, I'm 69% in fixed right now, and I plan on staying heavy into fixed now that I'm in the withdrawal years. But, it's still an interesting question based on the data points in the study.
Agreed. If I could reprogram my brain to have a higher tolerance for risk, I most likely would be better off. Alas, it seems to be an impossible task for me to totally tune out the barrage of daily noise that sends me scurrying to check account balances and worrying about depleting my IRA.

Is risk tolerance hardwired in us because of genetics or perhaps personal history?

ResNullius
Posts: 2091
Joined: Wed Oct 24, 2007 3:22 pm

Post by ResNullius » Sat Apr 02, 2011 8:45 am

Unless I've misread it, it seems that the past ten/eleven years have had little or no effect on the original study's findings and conclucsions.

User avatar
grayfox
Posts: 4934
Joined: Sat Sep 15, 2007 4:30 am

Post by grayfox » Sat Apr 02, 2011 9:06 am

OK, I am actually reading the words in the paper and not just the tables.

In one paragraph under "Data and Methodology", the authors discuss why they think historical using rolling periods is superior to Monte Carlo:
Cooley, et al wrote:The overlapping or rolling periods methodology has advantages over the commonly used Monte Carlo simulation and bootstrapping methodologies. Both Monte Carlo simulation and bootstrapping techniques usually assume that the mean and standard deviation of returns to securities and portfolios are unchanged through a payout period. However, analyses of stock and bond returns show varying means and standard deviations over time. For example, Cooley, Hubbard, and Walz (2003a) demonstrate the equivalency of portfolio success rates from Monte Carlo simulation versus rolling periods, but only if the returns data in the Monte Carlo analysis are adjusted for mean reversion and serial correlation. Simple simulation studies ignore serial correlation and clustered variances of stock returns over time. Simulation studies of withdrawal rates can be especially useful when the returns data are limited. Some examples are hedge funds and commodity and foreign stock ETFs. In those instances, there are too few years of returns data to conduct a traditional rolling periods study. Monte Carlo simulation and bootstrap studies can provide interesting results, but the rolling periods approach has the unique advantage of retaining the effects of the actual sequence of security returns and variances in bull markets and bear markets over the 84 years of data we researched.
The three main problems with Monte Carlo that they point out:
--parameters like mean and standard deviation vary over time
--must include mean reversion and serial correlation in models
--random return sequences loses the important effect of actual sequence of returns in bull and bear markets

I think there are others like Jim Otar who also used Monte Carlo simulation and later rejected it and are in favor of using historical data.

I've come to the conclusion that Monte Carlo simulation is only as good as the models that you use. Garbage-in, garbage out.

User avatar
grayfox
Posts: 4934
Joined: Sat Sep 15, 2007 4:30 am

Post by grayfox » Sat Apr 02, 2011 9:22 am

In another paragraph where they discuss Table 4 which shows median end values for inflation-adjusted withdrawals, they conclude that high stock allocations are better than low stock allocations. Apparently, the higher the better all the way up to 100% stock.
Cooley, et al wrote:Table 4 presents the median ending portfolio values per $1,000 initial retirement portfolio with inflation-adjusted withdrawal rates. Again, the benefit of a higher allocation to stocks is striking. Assuming a 6 percent real withdrawal rate and 30-year retirement period, portfolios consisting of 50 percent stocks and 50 percent bonds have a median end-of-period value of $9. Holding all else constant, the median ending value of portfolios with a 100 percent investment in equities is equal to $4,128. Thus, the median values in Table 4 support the notion that portfolios with a high equity composition are more supportive of inflation-adjusted withdrawals than portfolios with less equity investment.

Locked