A different theoretical view of SWR

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Ron Scott
Posts: 816
Joined: Tue Apr 05, 2016 5:38 am

A different theoretical view of SWR

Post by Ron Scott » Sat Jun 16, 2018 11:21 am

If you're like me and believe that churning calculations of historical market returns is not a useful exercise for predicting future returns, you're still left with everyone's problem: how to determine a SWR.

From a theoretical perspective I've come to believe that over the long term reasonable investments in stocks and bonds ultimately serve to accomplish one major feat. They help a portfolio combat inflation.

If we assume the affect of inflation is eliminated through investing we can develop a SWR by simply dividing the portfolio by the lifespan its needed to support. This approach is clean and simple. It obviously DOES NOT eliminate market risk during a retirement and it will never yield a perpetual portfolio. It simply replaces the need to estimate returns based on old market data and "professional" guesswork with the theory that the primary result of thoughtful investing is beating the inflation risk.

Given a 30-year retirement period this yields a SWR of 3.33%. SWRs can easily be calculated for any period and a personal SWR can be developed by assuming a longer or shorter lifespan, planning for one-off expenses or heirs, etc.

Thoughts?

tibbitts
Posts: 7944
Joined: Tue Feb 27, 2007 6:50 pm

Re: A different theoretical view of SWR

Post by tibbitts » Sat Jun 16, 2018 11:30 am

Ron Scott wrote:
Sat Jun 16, 2018 11:21 am
If you're like me and believe that churning calculations of historical market returns is not a useful exercise for predicting future returns, you're still left with everyone's problem: how to determine a SWR.

From a theoretical perspective I've come to believe that over the long term reasonable investments in stocks and bonds ultimately serve to accomplish one major feat. They help a portfolio combat inflation.

If we assume the affect of inflation is eliminated through investing we can develop a SWR by simply dividing the portfolio by the lifespan its needed to support. This approach is clean and simple. It obviously DOES NOT eliminate market risk during a retirement and it will never yield a perpetual portfolio. It simply replaces the need to estimate returns based on old market data and "professional" guesswork with the theory that the primary result of thoughtful investing is beating the inflation risk.

Given a 30-year retirement period this yields a SWR of 3.33%. SWRs can easily be calculated for any period and a personal SWR can be developed by assuming a longer or shorter lifespan, planning for one-off expenses or heirs, etc.

Thoughts?
So for the typical 35-40yr retirement you're at the same 2.5-3%-ish number that's becoming the average expectation around here, so it's the same result you're just getting there via a different means. We still hear about the extreme 1% and 4% positions but most people are settling for something in the middle.

liberty53
Posts: 33
Joined: Sat Oct 18, 2014 7:31 am
Location: Boston Area

Re: A different theoretical view of SWR

Post by liberty53 » Sat Jun 16, 2018 11:33 am

Well - 3.33% is less than 4% so already very conservative.

Do you adjust the withdrawal to account for inflation - or does it stay fixed?

Do you recalculate every year? Do I get to take 5% 10 years into my 30 year retirement?

Ron Scott
Posts: 816
Joined: Tue Apr 05, 2016 5:38 am

Re: A different theoretical view of SWR

Post by Ron Scott » Sat Jun 16, 2018 12:25 pm

liberty53 wrote:
Sat Jun 16, 2018 11:33 am
Well - 3.33% is less than 4% so already very conservative.

Do you adjust the withdrawal to account for inflation - or does it stay fixed?

Do you recalculate every year? Do I get to take 5% 10 years into my 30 year retirement?
Whoa...

1. This is not well-formed theory. This is just the beginning, based on skepticism around using market performance history to predict the future and a general belief that investing has the inherent power to negate the effects of inflation. I'm trying to get at the problem without using the same historical databases others use.

2. 3.33% is neither "very conservative" or aggressive. It's just the simple math. Those who call it conservative are comparing it to the history-based approach (The 4% Rule), which I don't believe in. (And you call it "very conservative" while tibbitts above calls my number the norm and yours extreme, so...)

3. So far I don't see the need to recalculate but I understand the argument. My initial thought is that investing combats inflation over the long term and should not be tweaked all the time as the "combat experience" can be fluid. You can win the war (over your lifetime) while losing some battles (annually).

At this time I'm more interested in the debate than in defending the theory.

MotoTrojan
Posts: 2121
Joined: Wed Feb 01, 2017 8:39 pm

Re: A different theoretical view of SWR

Post by MotoTrojan » Sat Jun 16, 2018 12:33 pm

To me this works best on shorter time periods as it gets overly conservative for longer durations.

What portfolio makeup would go best with this? TIPs?

User avatar
nisiprius
Advisory Board
Posts: 36499
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: A different theoretical view of SWR

Post by nisiprius » Sat Jun 16, 2018 12:39 pm

It is often overlooked--thanks to "spin" by people whose incomes depending on sales of stocks and bonds, and a too-strict interpretation of the word "cash"--that (assuming reasonable due diligence in selection),

...bank savings accounts, CDs, and Treasury bills have historically kept up with inflation and slightly more--they've provided a small positive real return, with emphasis both on "small" and "positive." Ditto series I savings bonds, and TIPs.

So, 3.33%/year for a 30-year retirement period works for these, too.

In a sense the argument is over. Back in the 1990s people were claiming you could withdraw 7-8%. The big discovery is that you can't.

The SWR cannot be anything more than a ballpark number, and the ballpark is 3-4%. You can't pin it down any closer than that, you can't have any absolute assurance that any number will work.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.

tibbitts
Posts: 7944
Joined: Tue Feb 27, 2007 6:50 pm

Re: A different theoretical view of SWR

Post by tibbitts » Sat Jun 16, 2018 12:43 pm

liberty53 wrote:
Sat Jun 16, 2018 11:33 am
Well - 3.33% is less than 4% so already very conservative.

Do you adjust the withdrawal to account for inflation - or does it stay fixed?

Do you recalculate every year? Do I get to take 5% 10 years into my 30 year retirement?
As far as I know the 4% number was never claimed to be valid for a global portfolio which virtually every Boglehead implements today.

User avatar
JoMoney
Posts: 5591
Joined: Tue Jul 23, 2013 5:31 am

Re: A different theoretical view of SWR

Post by JoMoney » Sat Jun 16, 2018 12:49 pm

Yup. But you'll still find people on this board who believe they need 50x expenses to fund 30 years :shock:
When I've looked at withdrawals rates, basing it on life expectancy makes the most sense to me. In the later years you can likely use remaining balance to buy an immediate annuity to insure you don't run out of money.
"To achieve satisfactory investment results is easier than most people realize; to achieve superior results is harder than it looks." - Benjamin Graham

dbr
Posts: 27207
Joined: Sun Mar 04, 2007 9:50 am

Re: A different theoretical view of SWR

Post by dbr » Sat Jun 16, 2018 1:11 pm

Ron Scott wrote:
Sat Jun 16, 2018 11:21 am

From a theoretical perspective I've come to believe that over the long term reasonable investments in stocks and bonds ultimately serve to accomplish one major feat. They help a portfolio combat inflation.

Given a 30-year retirement period this yields a SWR of 3.33%. SWRs can easily be calculated for any period and a personal SWR can be developed by assuming a longer or shorter lifespan, planning for one-off expenses or heirs, etc.

Thoughts?
From that assumption you get that result. From a different assumption you get a different result. At this moment the 30 year TIPS real yield is about 1%. Barring some practical details a ladder of 30 years TIPS would support a 3.8% inflation indexed SWR with no uncertainty. I guess I am not impressed.

liberty53
Posts: 33
Joined: Sat Oct 18, 2014 7:31 am
Location: Boston Area

Re: A different theoretical view of SWR

Post by liberty53 » Sat Jun 16, 2018 1:14 pm

Ron Scott wrote:
Sat Jun 16, 2018 12:25 pm
liberty53 wrote:
Sat Jun 16, 2018 11:33 am
Well - 3.33% is less than 4% so already very conservative.

Do you adjust the withdrawal to account for inflation - or does it stay fixed?

Do you recalculate every year? Do I get to take 5% 10 years into my 30 year retirement?
Whoa...

1. This is not well-formed theory. This is just the beginning, based on skepticism around using market performance history to predict the future and a general belief that investing has the inherent power to negate the effects of inflation. I'm trying to get at the problem without using the same historical databases others use.

2. 3.33% is neither "very conservative" or aggressive. It's just the simple math. Those who call it conservative are comparing it to the history-based approach (The 4% Rule), which I don't believe in. (And you call it "very conservative" while tibbitts above calls my number the norm and yours extreme, so...)

3. So far I don't see the need to recalculate but I understand the argument. My initial thought is that investing combats inflation over the long term and should not be tweaked all the time as the "combat experience" can be fluid. You can win the war (over your lifetime) while losing some battles (annually).

At this time I'm more interested in the debate than in defending the theory.
Well - you say that the portfolio should keep pace with inflation - so what do you do with the portfolio? I asked if you adjust withdrawals because I believe you were implying an adjustment based on the portfolio gains. If you don't adjust withdrawals then you obviously will be reducing standard of living due to inflation eating into the purchasing power of the "fixed" withdrawal amounts.

The benchmark (whether we like it or not) is 4%. If I understand this correctly - the 4% was a "round down" from a roughly 4.5% SWR determined by Bengen. That's why I called 3.3% conservative. The 3.3% WR is 20% more conservative than the 4% WR .

Snowjob
Posts: 1488
Joined: Sun Jun 28, 2009 10:53 pm

Re: A different theoretical view of SWR

Post by Snowjob » Sat Jun 16, 2018 1:19 pm

So if a 4% swr gets me around a 95% success rate, and ill probabaly only have maybe a 60% chance of making it to the end anyway that seems like fairly good odds 4% is a reasonable swr to use. Throw in the ability to add anuities, cut back a little during lean years and the observed natiral decline in spending as one ages and 4% starts looking even better. Bounce all this off of bernstiens suggestion that anyting over 80% chance of success is a meaningless calculation and im pretty certain 4% is the way to plan

Church Lady
Posts: 459
Joined: Sat Jun 28, 2014 7:49 pm

Re: A different theoretical view of SWR

Post by Church Lady » Sat Jun 16, 2018 6:03 pm

https://www.bogleheads.org/wiki/Withdra ... al_amounts

Are you basically proposing 1/N withdrawal amounts as described on this page?

A weakness with this method is that we don't really know how long we
will live. Talk about churning historical numbers!

A bigger problem is that 1/30 is way below the perpetual withdrawal rate (PWR) for a 60/40 portfolio.
For a very long retirement, say 40 years for a 60 y.o. retiring today, the sequence is 2.5%, 2.56%, 2.6%, etc. This too is far below the PWR for a 60/40 porfolio.

Click this link https://portfoliocharts.com/2016/12/09/ ... etirement/ to see the PWR for a 60/40 portfolio.

(If you have a problem with the PWR as calculated on that page, please post your reasons as I am using that page to help plan my retirement spending. Thanks.)

So back to early year withdrawals.
You're leaving :moneybag on the table precisely when you are young enough to enjoy spending it.
A PWR would be more money to spend, and it would protect your stash for your heirs or an
unexpectedly long life.

Am I missing something here? If I misinterpreted your scheme, my apologies.
He that loveth silver shall not be satisfied with silver; nor he that loveth abundance with increase: this is also vanity. Ecclesiastes 1:8

User avatar
Sandtrap
Posts: 5252
Joined: Sat Nov 26, 2016 6:32 pm
Location: Hawaii😀 Northern AZ.😳 Retired.

Re: A different theoretical view of SWR

Post by Sandtrap » Sat Jun 16, 2018 6:20 pm

At a minimum, keeping up with inflation makes sense if looked at this way. (albeit simplistically). :shock:
For example:
30 year projection post retirement.
Annual expenses 100k/year.
3 million in the portfolio in any mix and proportion.
Or. . . in a mattress or pile in a "storage unit" such as in the movie "Breaking Bad'. :D
That would work.
Any yield beyond keeping up with inflation would be a bonus for the hares.
j

User avatar
Sandtrap
Posts: 5252
Joined: Sat Nov 26, 2016 6:32 pm
Location: Hawaii😀 Northern AZ.😳 Retired.

Re: A different theoretical view of SWR

Post by Sandtrap » Sat Jun 16, 2018 6:23 pm

JoMoney wrote:
Sat Jun 16, 2018 12:49 pm
Yup. But you'll still find people on this board who believe they need 50x expenses to fund 30 years :shock:
When I've looked at withdrawals rates, basing it on life expectancy makes the most sense to me. In the later years you can likely use remaining balance to buy an immediate annuity to insure you don't run out of money.
+1
Makes perfect sense.
j

jalbert
Posts: 3588
Joined: Fri Apr 10, 2015 12:29 am

Re: A different theoretical view of SWR

Post by jalbert » Sat Jun 16, 2018 10:13 pm

tibbitts wrote:
Sat Jun 16, 2018 12:43 pm
liberty53 wrote:
Sat Jun 16, 2018 11:33 am
Well - 3.33% is less than 4% so already very conservative.

Do you adjust the withdrawal to account for inflation - or does it stay fixed?

Do you recalculate every year? Do I get to take 5% 10 years into my 30 year retirement?
As far as I know the 4% number was never claimed to be valid for a global portfolio which virtually every Boglehead implements today.
I think Wade Pfau did SWR simulations using a portfolio that was 50% equities and 50% bonds, each split half-and-half between US and non-US assets.

My memory of the findings is that the inflationary period of the 1970’s was more challenging for the all-US portfolio and the deflationary period of the 1920’s was more challenging for the globally diversified portfolio.
Index fund investor since 1987.

JoinToday
Posts: 754
Joined: Sat Mar 10, 2007 9:59 pm

Re: A different theoretical view of SWR

Post by JoinToday » Sat Jun 16, 2018 10:38 pm

Look at the method proposed by Henry Hebler, he called the method "autopilot".
https://www.marketwatch.com/story/put-r ... genumber=2

I am planning on doing something like what he proposes. You can withdraw 1/N (%) of your portfolio each year. Normally "N" would be reduced by 1 for each additional year, but Hebeler uses an IRS RMD schedule, where N is reduced by a little less than 1 for each additional year. This effectively stretches out your portfolio so you don't run out of money beyond your date of death predicted at the beginning of your retirement.

In addition, he includes 2 year smoothing in the withdrawal.

You can fairly easily modify his method to increase smoothing, and change the longevity.
I wish I had learned about index funds 25 years ago

jalbert
Posts: 3588
Joined: Fri Apr 10, 2015 12:29 am

Re: A different theoretical view of SWR

Post by jalbert » Sat Jun 16, 2018 11:12 pm

Ron Scott wrote:
Sat Jun 16, 2018 11:21 am
If you're like me and believe that churning calculations of historical market returns is not a useful exercise for predicting future returns, you're still left with everyone's problem: how to determine a SWR.

From a theoretical perspective I've come to believe that over the long term reasonable investments in stocks and bonds ultimately serve to accomplish one major feat. They help a portfolio combat inflation.

If we assume the affect of inflation is eliminated through investing we can develop a SWR by simply dividing the portfolio by the lifespan its needed to support. This approach is clean and simple. It obviously DOES NOT eliminate market risk during a retirement and it will never yield a perpetual portfolio. It simply replaces the need to estimate returns based on old market data and "professional" guesswork with the theory that the primary result of thoughtful investing is beating the inflation risk.

Given a 30-year retirement period this yields a SWR of 3.33%. SWRs can easily be calculated for any period and a personal SWR can be developed by assuming a longer or shorter lifespan, planning for one-off expenses or heirs, etc.

Thoughts?
If you are only expecting to keep up with inflation with your stock and bond portfolio, why would you take the risk instead of just throwing it all into TIPS?
Index fund investor since 1987.

smectym
Posts: 188
Joined: Thu May 26, 2011 5:07 pm

Re: A different theoretical view of SWR

Post by smectym » Sat Jun 16, 2018 11:26 pm

A little-discussed problem with SWR theory is that it always assumes a unitary portfolio of x stocks / y bonds, consisting of substantially all a retiree’s investable assets, all of which is being drawn down at this or that % rate. These are simplifying assumptions convenient for facilitating academic debate and enabling apples-to-apples comparisons ..but perhaps not the best model for any individual retiree or retiree household to follow.

In our case, we’re planning to draw down certain elements of the overall portfolio (chiefly the predictable IRA/401(k) pieces) while leaving other puzzle pieces to grow for a while (e.g., savings bonds, variable annuity, taxable portfolio).

Without derailing the discussion (and yet in a way introducing such wrinkles DOES inevitably derail the discussion by transgressing traditional SWR discussion ground rules), just want to point out that for some, perhaps many retirees, a more staggered approach to decumulation may lower certain risks (e.g. “running out of money”) and perhaps optimize long-term portfolio returns (bc certain assets are not subject to drawdown). More difficult to “prove” any of this—but has the research been done to disprove it?

Smectym

AlohaJoe
Posts: 3688
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: A different theoretical view of SWR

Post by AlohaJoe » Sun Jun 17, 2018 12:20 am

smectym wrote:
Sat Jun 16, 2018 11:26 pm
A little-discussed problem with SWR theory is that it always assumes a unitary portfolio of x stocks / y bonds, consisting of substantially all a retiree’s investable assets, all of which is being drawn down at this or that % rate
This isn't true. I've seen plenty of SWR analyses that include things like human capital, Social Security, housing equity, pensions, ability to go back to work, etc.

IlliniDave
Posts: 2294
Joined: Fri May 17, 2013 7:09 am

Re: A different theoretical view of SWR

Post by IlliniDave » Sun Jun 17, 2018 6:30 am

That's a reasonable way of looking at it. I too am a little leery of taking too much stock either in history, or in forecasts. I also view withdrawal rate as just a first-order approximation and really don't intend to follow the withdrawal strategy that is usually linked with them. It's difficult to find a level one can declare "safe". I'm hoping to get in a position where I can enjoy a modest but comfortable lifestyle while maintaining the flexibility to spend a little more on occasion if the need appears or it strikes my fancy, and simultaneously have the flexibility to tighten my belt a little if need be without subjecting myself to abject misery and deprivation. So I put a little margin in what I plan for a spending baseline, and a little margin in asset sizing relative to that baseline, and will have to make a leap of faith that I'm able to make it work.
Don't do something. Just stand there!

Ron Scott
Posts: 816
Joined: Tue Apr 05, 2016 5:38 am

Re: A different theoretical view of SWR

Post by Ron Scott » Sun Jun 17, 2018 6:53 am

One benefit of the “keep up with inflation” theory is that it is NOT based on historical data. It is another attempt to address the same problem, by other means. That it yields a SWR somewhat close to what the historical-data theorists find should be reassuring.

I also find it interesting that there is a good diversity of opinion on this board regarding the final SWR itself. Some see 4% as conservative, some believe it’s right on target, and some think 3% is “safe” today. The proposed 3.33% (or n / 1 as some call it) stands up well in this pack.

Probably the most concerning comment on the subject is fear of “leaving money on the table”; too much obsession with FoMO, gambling and death for my tastes.if everyone left a little on the table the world would better for it.

User avatar
jimmyq
Posts: 155
Joined: Wed Apr 17, 2013 6:34 pm

Re: A different theoretical view of SWR

Post by jimmyq » Sun Jun 17, 2018 6:57 am

Ron Scott wrote:
Sat Jun 16, 2018 11:21 am
If we assume the affect of inflation is eliminated through investing we can develop a SWR by simply dividing the portfolio by the lifespan its needed to support. This approach is clean and simple.
While this approach is simple and clean, I think it is way over-simplified, especially for very long and very short retirements. Someone planning on retiring very early in their 40s that is looking at a 50 year retirement horizon would have a SWR of only 2%. Way too low unless future history looks nothing like the past. On the other hand, let's say you are a workaholic that (finally) retires at the age of 78, and you estimate that you have only 10 years left based on family history. Would you really be willing to stick to the plan and take out 10% every year? Probably not.

User avatar
JoMoney
Posts: 5591
Joined: Tue Jul 23, 2013 5:31 am

Re: A different theoretical view of SWR

Post by JoMoney » Sun Jun 17, 2018 7:04 am

jimmyq wrote:
Sun Jun 17, 2018 6:57 am
... at the age of 78, and you estimate that you have only 10 years left based on family history. Would you really be willing to stick to the plan and take out 10% every year? Probably not.
If it were me, and I reached that point and I expected to need 10% of my portfolio for annual expenses I would likely buy an immediate annuity. If the portfolio had remained large enough where I had a big margin of safety, probably pass on the annuity.
"To achieve satisfactory investment results is easier than most people realize; to achieve superior results is harder than it looks." - Benjamin Graham

cjking
Posts: 1801
Joined: Mon Jun 30, 2008 4:30 am

Re: A different theoretical view of SWR

Post by cjking » Sun Jun 17, 2018 7:28 am

A weakness with this method is that we don't really know how long we will live.
I would make "n" current actuarial life expectancy, and recalculate each year. Once life-expectancy falls below 15 years, if it looked at that point like finances are going to be tight, I would start converting to annuities.

The 1/n in effect assumes a 0% real return, which is as arbitrary as and not necessarily better than any other fixed number. I would suggest 50% of portfolio expected return as the anticipated return, then use PMT function parameterised with n and anticipated return. An alternative to trying to calculate expected returns would be to just use a fixed value of 2% as the anticipated return.

(I have in the past seen UK investment linked annuities where you could tilt the income stream forwards or backwards by choosing an anticipated rate of return, and 2% would have been a fairly central value to have chosen.)

cjking
Posts: 1801
Joined: Mon Jun 30, 2008 4:30 am

Re: A different theoretical view of SWR

Post by cjking » Sun Jun 17, 2018 7:38 am

So a worked example
Current balance: $1 million
Portfolio: 60% world index, (E10/P)-[fund charges] = 3.5%, 40% bond fund guesstimated return 1% real.
Expected real return: 60%*3.5%+40%*1% = 2.5%
Anticipated real return: 50% * 2.5% = 1.25%
Current actuarial life expectancy: 30 years
This years income: PMT(1.25%,30,-1000000,0,1) = $39,683.

Repeat the calculation each year with updated values for current balance, anticipated return and life expectancy.

cherijoh
Posts: 4932
Joined: Tue Feb 20, 2007 4:49 pm
Location: Charlotte NC

Re: A different theoretical view of SWR

Post by cherijoh » Sun Jun 17, 2018 8:07 am

Ron Scott wrote:
Sat Jun 16, 2018 11:21 am
If you're like me and believe that churning calculations of historical market returns is not a useful exercise for predicting future returns, you're still left with everyone's problem: how to determine a SWR.

From a theoretical perspective I've come to believe that over the long term reasonable investments in stocks and bonds ultimately serve to accomplish one major feat. They help a portfolio combat inflation.

If we assume the affect of inflation is eliminated through investing we can develop a SWR by simply dividing the portfolio by the lifespan its needed to support. This approach is clean and simple. It obviously DOES NOT eliminate market risk during a retirement and it will never yield a perpetual portfolio. It simply replaces the need to estimate returns based on old market data and "professional" guesswork with the theory that the primary result of thoughtful investing is beating the inflation risk.

Given a 30-year retirement period this yields a SWR of 3.33%. SWRs can easily be calculated for any period and a personal SWR can be developed by assuming a longer or shorter lifespan, planning for one-off expenses or heirs, etc.

Thoughts?
Personally, I think of the 4% rule or any other SWR rule as being most useful in the early stages of retirement planning (e.g., am I saving enough to be able to retire 20 years from now) and least useful when you are actually retired. By then you will know your actual fixed and discretionary expenses and will be able to adjust your annual withdrawal rate based on market conditions.

However, I will point out one fallacy of your proposed approach. You have deterministically set your retirement as a 30-year (or some other pre-specified period). No one knows how long they will live, therefore no one knows how long they will be retired. The well-documented 4% SWR study has the benefit of allowing you to determine the probability of your portfolio lasting across a variety of retirement periods and across a range of portfolio compositions.

User avatar
jimmyq
Posts: 155
Joined: Wed Apr 17, 2013 6:34 pm

Re: A different theoretical view of SWR

Post by jimmyq » Sun Jun 17, 2018 8:30 am

cjking wrote:
Sun Jun 17, 2018 7:28 am
... then use PMT function parameterised with n and anticipated return.
Yeah, I agree. It seems like anytime I try to come up with a "better" or "simpler" method, I always come back to some variation that uses the PMT function. Not quite so simple, but really not that difficult.

Horsefly
Posts: 264
Joined: Sat Oct 24, 2015 8:13 am
Location: Colorado, mostly

Re: A different theoretical view of SWR

Post by Horsefly » Sun Jun 17, 2018 10:52 am

cjking wrote:
Sun Jun 17, 2018 7:28 am
A weakness with this method is that we don't really know how long we will live.
I would make "n" current actuarial life expectancy, and recalculate each year. Once life-expectancy falls below 15 years, if it looked at that point like finances are going to be tight, I would start converting to annuities.

The 1/n in effect assumes a 0% real return, which is as arbitrary as and not necessarily better than any other fixed number. I would suggest 50% of portfolio expected return as the anticipated return, then use PMT function parameterised with n and anticipated return. An alternative to trying to calculate expected returns would be to just use a fixed value of 2% as the anticipated return.

(I have in the past seen UK investment linked annuities where you could tilt the income stream forwards or backwards by choosing an anticipated rate of return, and 2% would have been a fairly central value to have chosen.)
I use the PMT() spreadsheet function too, but mainly to establish a high water mark of what I can withdraw in a given year. So far (5 years of retirement) I've been well under the PMT result. What I like about this is that you can recalculate each year (or even more often if desired), adjusting both the end year and return based on health and whatever guess you want to make on the markets.

ncbill
Posts: 348
Joined: Sun Jul 06, 2008 4:03 pm

Re: A different theoretical view of SWR

Post by ncbill » Sun Jun 17, 2018 11:13 am

Isn't OP's approach similar to the "wasting asset" retirement model?

Ben Mathew
Posts: 96
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle
Contact:

Re: A different theoretical view of SWR

Post by Ben Mathew » Sun Jun 17, 2018 11:26 am

cjking wrote:
Sun Jun 17, 2018 7:38 am
So a worked example
Current balance: $1 million
Portfolio: 60% world index, (E10/P)-[fund charges] = 3.5%, 40% bond fund guesstimated return 1% real.
Expected real return: 60%*3.5%+40%*1% = 2.5%
Anticipated real return: 50% * 2.5% = 1.25%
Current actuarial life expectancy: 30 years
This years income: PMT(1.25%,30,-1000000,0,1) = $39,683.

Repeat the calculation each year with updated values for current balance, anticipated return and life expectancy.
This approach, based on current estimates of expected real returns, is much more appealing to me than using SWRs calculated from historical returns which occurred when valuations were much lower. There has to be some sort of adjustment for the fact that valuations are much higher now than was typical in the past. Otherwise we're not using all the information we have to plan for the future.

We could expand the above analysis to allow for changing asset allocations over time. That requires building a spreadsheet model rather than using the PMT function. The spreadsheet I use for this is here.

User avatar
siamond
Posts: 4168
Joined: Mon May 28, 2012 5:50 am

Re: A different theoretical view of SWR

Post by siamond » Sun Jun 17, 2018 11:28 am

cjking wrote:
Sun Jun 17, 2018 7:38 am
So a worked example
Current balance: $1 million
Portfolio: 60% world index, (E10/P)-[fund charges] = 3.5%, 40% bond fund guesstimated return 1% real.
Expected real return: 60%*3.5%+40%*1% = 2.5%
Anticipated real return: 50% * 2.5% = 1.25%
Current actuarial life expectancy: 30 years
This years income: PMT(1.25%,30,-1000000,0,1) = $39,683.

Repeat the calculation each year with updated values for current balance, anticipated return and life expectancy.
I do something similar, including the expected returns model, but:
- I would be more conservative about life expectancy, the actuarial life expectancy is a probabilistic midpoint, and you don't want to expose yourself to longevity risk (notably if you're healthy and have parents or grandparents who went in their 90s...)
- I don't see the point of slashing expected returns by a factor of 2. That is being unnecessarily conservative. The PMT function does work best with expected returns in the statistical sense, a probabilistic midpoint, and adjusts itself nicely in case of crises. No need to play it 'SWR-like', in other words.

My worked example would be... Ah funny, basically the same result!! :D :beer

Current balance: $1 million
Portfolio: 60% world index, (E10/P)-[fund charges] = 3.5%, 40% bond fund guesstimated return 1% real.
Expected real return: 60%*3.5% + 40%*1% = 2.5%
Conservative life expectancy: 30+10 = 40 years
This years income: PMT(2.5%, 40, -1000000, 0, 1) = $38,865.
Last edited by siamond on Sun Jun 17, 2018 11:29 am, edited 1 time in total.

EddyB
Posts: 534
Joined: Fri May 24, 2013 3:43 pm

Re: A different theoretical view of SWR

Post by EddyB » Sun Jun 17, 2018 11:28 am

Ron Scott wrote:
Sat Jun 16, 2018 12:25 pm
liberty53 wrote:
Sat Jun 16, 2018 11:33 am
Well - 3.33% is less than 4% so already very conservative.

Do you adjust the withdrawal to account for inflation - or does it stay fixed?

Do you recalculate every year? Do I get to take 5% 10 years into my 30 year retirement?
Whoa...

2. 3.33% is neither "very conservative" or aggressive. It's just the simple math. Those who call it conservative are comparing it to the history-based approach (The 4% Rule), which I don't believe in. (And you call it "very conservative" while tibbitts above calls my number the norm and yours extreme, so...)

At this time I'm more interested in the debate than in defending the theory.
You’ve made an assumption that a portfolio will only keep up with inflation, which is very conservative vs. a number of portfolios over prolonged periods of history. To then look at the second step and claim it’s “just math” is delusional (I don’t mean that it any insulting sense, I just mean you can’t ignore that you’re only doing that math on the product of a particular assumption and pretending that the input is a fundamental fact).

User avatar
siamond
Posts: 4168
Joined: Mon May 28, 2012 5:50 am

Re: A different theoretical view of SWR

Post by siamond » Sun Jun 17, 2018 11:44 am

cherijoh wrote:
Sun Jun 17, 2018 8:07 am
However, I will point out one fallacy of your proposed approach. You have deterministically set your retirement as a 30-year (or some other pre-specified period). No one knows how long they will live, therefore no one knows how long they will be retired.
That is exactly right. Those so-called LMP approaches, centered on spending a portfolio invested in a super conservative manner (e.g. TIPS), are NOT safe at all... They are actually extremely risky imho. Once your portfolio is mostly spent (say 2/3rd of the way), you're way past a point of no-return if you would allow me a bad pun... :wink:

User avatar
siamond
Posts: 4168
Joined: Mon May 28, 2012 5:50 am

Re: A different theoretical view of SWR

Post by siamond » Sun Jun 17, 2018 11:52 am

jalbert wrote:
Sat Jun 16, 2018 10:13 pm
I think Wade Pfau did SWR simulations using a portfolio that was 50% equities and 50% bonds, each split half-and-half between US and non-US assets.

My memory of the findings is that the inflationary period of the 1970’s was more challenging for the all-US portfolio and the deflationary period of the 1920’s was more challenging for the globally diversified portfolio.
You are exactly right. Simba to the rescue, I just ran a similar test:

Image

User avatar
JoMoney
Posts: 5591
Joined: Tue Jul 23, 2013 5:31 am

Re: A different theoretical view of SWR

Post by JoMoney » Sun Jun 17, 2018 11:54 am

siamond wrote:
Sun Jun 17, 2018 11:44 am
cherijoh wrote:
Sun Jun 17, 2018 8:07 am
However, I will point out one fallacy of your proposed approach. You have deterministically set your retirement as a 30-year (or some other pre-specified period). No one knows how long they will live, therefore no one knows how long they will be retired.
That is exactly right. Those so-called LMP approaches, centered on spending a portfolio invested in a super conservative manner (e.g. TIPS), are NOT safe at all... They are actually extremely risky imho. Once your portfolio is mostly spent (say 2/3rd of the way), you're way past a point of no-return if you would allow me a bad pun... :wink:
Every SWR study has to pick some time period. Nobody says they're perfect, but whats the alternative? How do you decide what a reasonable amount of savings is for an undefined period of time?
"To achieve satisfactory investment results is easier than most people realize; to achieve superior results is harder than it looks." - Benjamin Graham

User avatar
siamond
Posts: 4168
Joined: Mon May 28, 2012 5:50 am

Re: A different theoretical view of SWR

Post by siamond » Sun Jun 17, 2018 12:17 pm

JoMoney wrote:
Sun Jun 17, 2018 11:54 am
siamond wrote:
Sun Jun 17, 2018 11:44 am
cherijoh wrote:
Sun Jun 17, 2018 8:07 am
However, I will point out one fallacy of your proposed approach. You have deterministically set your retirement as a 30-year (or some other pre-specified period). No one knows how long they will live, therefore no one knows how long they will be retired.
That is exactly right. Those so-called LMP approaches, centered on spending a portfolio invested in a super conservative manner (e.g. TIPS), are NOT safe at all... They are actually extremely risky imho. Once your portfolio is mostly spent (say 2/3rd of the way), you're way past a point of no-return if you would allow me a bad pun... :wink:
Every SWR study has to pick some time period. Nobody says they're perfect, but whats the alternative? How do you decide what a reasonable amount of savings is for an undefined period of time?
By not using a withdrawal method that quickly depletes your savings until the point of no-return (like the one the OP suggested). Trick is to spend what the market returns, with some smoothing process to avoid making income get overly hectic, and keeping a decent engine of growth (i.e. stocks) in one's portfolio. You can do that with Guyton-Klinger decision rules, you can do that with PMT() derivatives (keeping the #periods above a certain threshold), etc. It doesn't have to foresee an infinite lifespan, but certainly something significantly longer than actuarial life expectancies. My mother passed away at 98 last year. And I would never bet against the progress of medicine in a couple of decades...

CurlyDave
Posts: 685
Joined: Thu Jul 28, 2016 11:37 am

Re: A different theoretical view of SWR

Post by CurlyDave » Sun Jun 17, 2018 12:17 pm

Anything except the most conservative portfolio should be able to do more than just keep up with inflation.

If you look at the Trinity study, most of the portfolios with at least 50% equities have very substantial terminal values.

Trying to be ultra-conservative has a price.

jalbert
Posts: 3588
Joined: Fri Apr 10, 2015 12:29 am

Re: A different theoretical view of SWR

Post by jalbert » Sun Jun 17, 2018 2:18 pm

I can’t find the reference but a recent article J read by an actuary who studied the problem compatibles SWR, income annuities, and the US govt RMD algorithm used for IRAs and found the RMD algorithm to be optimal, using a balanced portfolio. This is just one study among many.
Index fund investor since 1987.

AlphaLess
Posts: 591
Joined: Fri Sep 29, 2017 11:38 pm

Re: A different theoretical view of SWR

Post by AlphaLess » Sun Jun 17, 2018 2:23 pm

Ron Scott wrote:
Sat Jun 16, 2018 11:21 am
...theory...
Given a 30-year retirement period this yields a SWR of 3.33%. SWRs can easily be calculated for any period and a personal SWR can be developed by assuming a longer or shorter lifespan, planning for one-off expenses or heirs, etc.

Thoughts?
Sure. But that leaves one big problem open:
- when can I know that I can retire?

Also, what if the portfolio returns, relative to inflation, look like this:

1. -1%,
2. -1%,
3. -1%,
4. -1%,
5. -1%
etc

Also, I would argue that under your model, *ANY* withdrawal rate works for *ANY* duration.
Why?
Simply because you will never hit zero. However, you will VERY QUICKLY hit spending levels that are miniscule.

Ron Scott
Posts: 816
Joined: Tue Apr 05, 2016 5:38 am

Re: A different theoretical view of SWR

Post by Ron Scott » Sun Jun 17, 2018 2:47 pm

CurlyDave wrote:
Sun Jun 17, 2018 12:17 pm
Anything except the most conservative portfolio should be able to do more than just keep up with inflation.

If you look at the Trinity study, most of the portfolios with at least 50% equities have very substantial terminal values.
But you are using historical market return data to justify your conclusion. What we’re trying to do here is to calculate a safe withdrawal rate without reliance on historical market data. For those who do not believe past returns are predictive of future returns over a retirement period, your approach fails by definition.

I find it more comforting to ask myself what is the relationship between market returns and underlying inflation? And to be sure, my approach is not free from beliefs either. I believe the capitalist system helps stock and bond returns sit on top of inflation. And to be “safe“ I assume returns simply “keep pace with inflation“.

The 3.33% safe withdrawal rate suggested by my approach yields an annual spend about 17% less than the traditional 4% rule. That appears to be conservative only when compared to models that use historical data.

marcopolo
Posts: 1056
Joined: Sat Dec 03, 2016 10:22 am

Re: A different theoretical view of SWR

Post by marcopolo » Sun Jun 17, 2018 4:52 pm

Ron Scott wrote:
Sun Jun 17, 2018 2:47 pm
CurlyDave wrote:
Sun Jun 17, 2018 12:17 pm
Anything except the most conservative portfolio should be able to do more than just keep up with inflation.

If you look at the Trinity study, most of the portfolios with at least 50% equities have very substantial terminal values.
But you are using historical market return data to justify your conclusion. What we’re trying to do here is to calculate a safe withdrawal rate without reliance on historical market data. For those who do not believe past returns are predictive of future returns over a retirement period, your approach fails by definition.

I find it more comforting to ask myself what is the relationship between market returns and underlying inflation? And to be sure, my approach is not free from beliefs either. I believe the capitalist system helps stock and bond returns sit on top of inflation. And to be “safe“ I assume returns simply “keep pace with inflation“.

The 3.33% safe withdrawal rate suggested by my approach yields an annual spend about 17% less than the traditional 4% rule. That appears to be conservative only when compared to models that use historical data.
You say you don't trust historical data. Ok, so what makes you believe that stocks and bonds will out pace inflation in the future? Historically, they have, but you don't like that answer. I don't know of any laws of physics that will make that necessarily true in the future. If you are saying you want to be more conservative than historical norms, OK i see no problem with that. But, pretending you are not relying on historical pattern to arrive at your 3.33% just seems like a rationalization.
Once in a while you get shown the light, in the strangest of places if you look at it right.

User avatar
siamond
Posts: 4168
Joined: Mon May 28, 2012 5:50 am

Re: A different theoretical view of SWR

Post by siamond » Sun Jun 17, 2018 5:15 pm

Ron Scott wrote:
Sun Jun 17, 2018 2:47 pm
What we’re trying to do here is to calculate a safe withdrawal rate without reliance on historical market data. For those who do not believe past returns are predictive of future returns over a retirement period, your approach fails by definition.

I find it more comforting to ask myself what is the relationship between market returns and underlying inflation? And to be sure, my approach is not free from beliefs either. I believe the capitalist system helps stock and bond returns sit on top of inflation. And to be “safe“ I assume returns simply “keep pace with inflation“.
Hm, I had not fully appreciated your intent based on your OP, apologies if I sidetracked the thread a tad in past posts. Ok, so you might not be much of a believer in backtesting, but surely, you can appreciate that if an assertion is clearly disproven by backtesting, this brings quite a hole in a given belief, right?

Here is a growth chart, in real (inflation-adjusted) terms, for various asset allocations, starting from 0% (total/US) stocks and 100% (total/US) bonds, and then increasing the stock exposure. What is glaring (and I admittedly cherry-picked the starting point) is that 0/100 and 10/100 were under water after *FOURTY* years, and even 20/80 was quite close to go back in the red in the same timeframe.

Image

Were you aware of that extremely troublesome period of 40 years? This is just a honest question to see if your belief was solidly grounded or not.

Now you might answer that the 30/70 (or anything with more stocks) would have done ok, thank you very much. Well, yes, but this is historical knowledge... Furthermore, in the UK, there was an even more terrifying issue with bonds not tracking inflation for a lifetime (70 years, if I recall). I didn't run the numbers to see if 30/70 would have held up in the UK for a regular 30 years retirement period against inflation, but... I am not sure.

What I am trying to say is that even the most conservative assertions ('beliefs') can be really challenged by history. So I am not too sure about your desire to act out of any historical knowledge. Or... you elect to go 100% TIPS, then by definition, you track inflation, but this isn't consistent with your wording ('the capitalist system helps stock and bond returns sit on top of inflation').

ncbill
Posts: 348
Joined: Sun Jul 06, 2008 4:03 pm

Re: A different theoretical view of SWR

Post by ncbill » Sun Jun 17, 2018 7:20 pm

Why not do the "rising glide path" approach?

IIRC, that starts at a recommended 30/70 stock/bond AA.

cjking
Posts: 1801
Joined: Mon Jun 30, 2008 4:30 am

Re: A different theoretical view of SWR

Post by cjking » Mon Jun 18, 2018 3:20 am

Ron Scott wrote:
Sun Jun 17, 2018 2:47 pm
I find it more comforting to ask myself what is the relationship between market returns and underlying inflation? And to be sure, my approach is not free from beliefs either. I believe the capitalist system helps stock and bond returns sit on top of inflation. And to be “safe“ I assume returns simply “keep pace with inflation“.
If I were thinking along similar lines to you for picking numbers I would pick 0% as my anticipated return for bonds and 2% for equities. I think in the region of 2% is the worst ever return of the S&P 500 over any 30 year period.

Whatever number you choose, there's some possibility things will turn out worse. So you need a strategy that adjusts income according to the returns you actually get.

Once you have a variable withdrawal strategy, the anticipated return becomes a lot less important. The anticipated return being too high just means your income will be higher at the start of retirement than the end, it won't cause disaster.

A good variable withdrawal strategy with no conservatism in the parameters is a better way to protect against extreme scenarios than fixed withdrawals with very conservative parameters.

(I think you are proposing fixed withdrawals of initial_balance/retirement_length. A very similar strategy would be current_balance/years_left, recalculated yearly. That would be roughly as good as PMT with positive anticipated returns, for periods of 30 years or less. For longer periods it starts to deliver an unreasonably low starting income, I believe.)

cjking
Posts: 1801
Joined: Mon Jun 30, 2008 4:30 am

Re: A different theoretical view of SWR

Post by cjking » Mon Jun 18, 2018 4:21 am

siamond wrote:
Sun Jun 17, 2018 11:28 am
I do something similar, including the expected returns model, but:
- I would be more conservative about life expectancy, the actuarial life expectancy is a probabilistic midpoint, and you don't want to expose yourself to longevity risk (notably if you're healthy and have parents or grandparents who went in their 90s...)
I like that use of actuarial life expectancy adjusts along the way. To make up some numbers, at 50 you might plan for 35 years, but by the time you are 85, you are planning for 10 years, not zero. Your life expectancy conditional on current age achieved does increase with time, so why not take that into account, just as you would take into account changing balances or changing expected returns.

I have a secondary aim of not running down the portfolio, which I hope to achieve by placing a cap on annual withdrawals. My strategy runs the portfolio down towards zero, in bad scenarios, where the cap has little or no effect. This serves the primary objective of delivering the best possible income stream. In other scenarios, the cap means I have a good chance of achieving my secondary objective, not running down the portfolio at all. (In fact, a lot of the time, the cap also delivers a third benefit: non-volatile income.)

As a consequence of the cap, it's more likely than not that my asset balance will not deplete during retirement, in which case there will be no need to worry about longevity. In much worse scenarios, and only in those scenarios, I would annuitise between ages 75 and 80, dealing with the longevity issue that way.

There's no scenario where I would be running the portfolio all the way down to zero while using PMT with actuarial life expectancy less than about 10 years.

- I don't see the point of slashing expected returns by a factor of 2. That is being unnecessarily conservative.
I don't believe the anticipated return matters much, within reason. I think it would be OK to use the full expected return without making any other changes, using half is just a personal preference to introduce one dimension of conservatism into the overall plan. I think generally half of expected return is the most conservative it's reasonable to be with anticipated return, except that for for periods of less than 30 years I don't think it matters if someone goes all the way done to zero, assuming they are OK with the starting income that gives.

I did once do an exercise with VPW where I defined a metric to be optimised, I think it was something like number of years with income below some very low threshold, and I found that using 60% of expected return was optimal for the metric, but also not vastly better than using other numbers.

My worked example would be... Ah funny, basically the same result!! :D :beer
I remember realising it made little difference to early years income, when switching the parameter to be conservative with from longevity to returns.

Valuethinker
Posts: 35683
Joined: Fri May 11, 2007 11:07 am

Re: A different theoretical view of SWR

Post by Valuethinker » Mon Jun 18, 2018 4:39 am

jalbert wrote:
Sat Jun 16, 2018 10:13 pm
tibbitts wrote:
Sat Jun 16, 2018 12:43 pm
liberty53 wrote:
Sat Jun 16, 2018 11:33 am
Well - 3.33% is less than 4% so already very conservative.

Do you adjust the withdrawal to account for inflation - or does it stay fixed?

Do you recalculate every year? Do I get to take 5% 10 years into my 30 year retirement?
As far as I know the 4% number was never claimed to be valid for a global portfolio which virtually every Boglehead implements today.
I think Wade Pfau did SWR simulations using a portfolio that was 50% equities and 50% bonds, each split half-and-half between US and non-US assets.

My memory of the findings is that the inflationary period of the 1970’s was more challenging for the all-US portfolio and the deflationary period of the 1920’s was more challenging for the globally diversified portfolio.
Just a sense check: did you mean 1930s?

There were some big fluctuations in inflation in the 1920s. US had very high inflation immediately post WW1, then deflation? UK something similar, followed by a virtual economic crash in 1924 (the year of the General Strike; the Minister of Finance, Winston Churchill, was advised to restore the Pound to the Gold Standard at its prewar rate of $4.85 to the pound. That led to serious deflation in the UK economy and an effort by coal mine owners (the largest category of industrial labour) to lower wages and increase working hours. Hence the strike in support of the miners. Churchill's role in suppressing that strike has not been forgotten).

(Germany is another story in that the currency collapsed during the hyperinflation that followed the French occupation of the Rhineland industrial area).

But, overall, I think prices were stable (or rose) at least until October, 1929?

Leroy Jones
Posts: 23
Joined: Thu Nov 16, 2017 8:14 pm

Re: A different theoretical view of SWR

Post by Leroy Jones » Mon Jun 18, 2018 5:43 am

Doesn't all the discussion about safe withdrawal rates for 30 or 40 years become mute once you reach 70.5 and the government starts mandating your withdrawal rate from your TIRA or 401K? At 70.5 you must withdraw .0365% and by 73 your at .0405% and it continues to increase from there regardless of what may be safe or not. With a 50% penalty for not taking the right amount the incentive is there to follow the law.

msk
Posts: 884
Joined: Mon Aug 15, 2016 10:40 am

Re: A different theoretical view of SWR

Post by msk » Mon Jun 18, 2018 6:35 am

I suspect that many people who discuss WR in detail already have more than enough to retire on. And are already retired. The query to myself (before I discovered this forum) was how high should my WR be so that I neither expose myself to ever running out of money no matter how long I live, and also to give away (to heirs and charity) as much as possible without jeopardizing my own, inflation-corrected, needs. I downloaded data for the 50 years 1966 to 2016, set up a spreadsheet, and started playing with WR strategies. It becomes very clear very quickly that with a highish WR you MUST react to market performance. A market collapse in one year should reduce your current cash withdrawal, regardless of inflation. I arrived at the simple strategy of withdrawing 5% of a 100% stocks portfolio annually. Using 50-year history we find that the remaining portfolio keeps up with inflation forever, and your cash annual withdrawals, despite yo-yoing with annual market performance, also keep up with inflation, forever. When I discovered BHs I used the MonteCarlo simulations to confirm the same as the Median forecast. 300 year history also corroborates that Trade and Industry (is this = stocks?) returned 5+% p.a. real terms. Any WR strategy that uses a % of annual, varying, portfolio value will survive forever., i.e. any rate less than 100%! The question is really how does one preserve the remaining REAL value? Answer, 5% p.a. My own needs are less than 5%, but I plan to give away the balance that gets to 5%. Market collapse? I will give away less, but keep the total close to 5%.

aristotelian
Posts: 4614
Joined: Wed Jan 11, 2017 8:05 pm

Re: A different theoretical view of SWR

Post by aristotelian » Mon Jun 18, 2018 7:31 am

Leroy Jones wrote:
Mon Jun 18, 2018 5:43 am
Doesn't all the discussion about safe withdrawal rates for 30 or 40 years become mute once you reach 70.5 and the government starts mandating your withdrawal rate from your TIRA or 401K? At 70.5 you must withdraw .0365% and by 73 your at .0405% and it continues to increase from there regardless of what may be safe or not. With a 50% penalty for not taking the right amount the incentive is there to follow the law.
No, when you withdraw from your 401k, you are just transferring from one account to another, less tax. You do not have to spend it. When SWR research is talking about withdrawal, it really means spending, where the money leaves your portfolio completely.

User avatar
siamond
Posts: 4168
Joined: Mon May 28, 2012 5:50 am

Re: A different theoretical view of SWR

Post by siamond » Mon Jun 18, 2018 7:51 am

cjking wrote:
Mon Jun 18, 2018 4:21 am
siamond wrote:
Sun Jun 17, 2018 11:28 am
I do something similar, including the expected returns model, but:
- I would be more conservative about life expectancy, the actuarial life expectancy is a probabilistic midpoint, and you don't want to expose yourself to longevity risk (notably if you're healthy and have parents or grandparents who went in their 90s...)
I like that use of actuarial life expectancy adjusts along the way. To make up some numbers, at 50 you might plan for 35 years, but by the time you are 85, you are planning for 10 years, not zero. Your life expectancy conditional on current age achieved does increase with time, so why not take that into account, just as you would take into account changing balances or changing expected returns. [...]
Oh yes, for sure, I certainly agree with that, it makes no sense to plan for #periods to go down to zero, and yes, it's good to use something adaptive along the way (or at least towards the end). I was just suggesting that "actuarial life expectancy + 10" (or something like that) might be a sounder way to proceed. You don't want to plan for an average life expectancy. You have a 50% chance of living longer than that, and then this would overly skew withdrawals towards the early years.

Now, of course, if you compensate by using a very conservative estimate of expected returns, this somehow balances out, but I don't know, I'd rather use each parameter for what it is truly intended. I think this will go a long way in sticking to the course.

Post Reply