Early retiree safe withdrawal done right?

[Totoro]

What's been bothering me about most SWR studies i've seen so far (specifically Bengen & Pfau) is that it is geared towards people with a rather limited future lifespan, with a typical maximum of 30 years.

This means that you can draw from two components derived from your savings:

1. Principal (the balance sheet)
2. Total real return from your portfolio (P&L)

For a (potential) very early retiree this makes those studies incomplete it seems.

Specifically: once you go below a withdrawal rate of roughly 3.5% I can guarantee you a 100% survival rate for 30 years: buy TIPS and you are done! For an early retiree tough, it would spell certain disaster after those 30 years. Early retirees cannot rely on the first component.

At the same time, it is clearly shown in those studies that the returns of the first 10 years practically fully determine whether you'll manage a long financially sound retirement.

So why not go for a different approach better suited for long-lived people and very early retirees?

For example: Look at how much *real* return I can expect in the coming 10 years?

Well, I did just that, and went one step further:I put it alongside the CAPE-10 (inverted) and low and behold, the following picture comes out (I'll show it in the next post, struggling to include it.)

The red arrow is at the current CAPE-10: 25. I did the math and it seems you'll only have a 20% chance of getting at least a real 4% return on stocks starting from valuations above this level

Note that this is just a preliminary analysis, using shiller numbers from 1929 until 2012 (roughly), and it does not reinvest dividends so it's slightly off. I can do something much better and complete if the forum finds this useful (and if it hasn't been done before). Let me know.

 

[Totoro]

Graph in attachment (I hope) - you can click on it to enlarge

Horizontal X-axis = 10-year real return in annual %, including dividends (not reinvested)
Vertical Y-Axis = Shiller CAPE-10 inverted at start date

 

[haha]

So why not go for a different approach better suited for long-lived people and very early retirees?

For example: Look at how much *real* return I can expect in the coming 10 years?

Well, I did just that, and went one step further:I put it alongside the CAPE-10 (inverted) and low and behold, the following picture comes out (I'll show it in the next post, struggling to include it.)

The red arrow is at the current CAPE-10: 25. I did the math and it seems you'll only have a 20% chance of getting at least a real 4% return on stocks starting from valuations above this level

Note that this is just a preliminary analysis, using shiller numbers from 1929 until 2012 (roughly), and it does not reinvest dividends so it's slightly off. I can do something much better and complete if the forum finds this useful (and if it hasn't been done before). Let me know.

This is sacrilege. Someone will be along shortly to arrest you and put you in the stocks.


Ha

 

[REWahoo]

The answer is obvious: keep working until you get within 30 years of your expiration date.

 

[razztazz]

Bengen actually calculated what I think he called a "Forever Withdrawal Rate" or something to that effect. It was actually designed to last 70 years the assumption being you'd have to work at least part of your life to get the money then you'd be over 100 when the money ran out so for all intents and purposes it was forever.

I believe his conclusion was 3.3%. This was on the order of 10-15 years ago. Sorry, I no longer have links to Bengen's articles on the subject. Perhaps this rings a bell with some of yous

 

[NW-Bound]

Graph in attachment (I hope) - you can click on it to enlarge

Horizontal X-axis = 10-year real return in annual %, including dividends (not reinvested)
Vertical Y-Axis = Shiller CAPE-10 inverted at start date

The descriptions of the axes appear swapped to me.

Anyway, equity return over long periods is a problem that I do not worry about much, not being so young. If I live for another 30 years, I suspect that at the end of that period I will not care much about money, having other things to think about (if I can even think clearly at that age).

 

[foxfirev5]

The answer is obvious: keep working until you get within 30 years of your expiration date.

I wish I could find that date stamped someplace. Meanwhile, like most I plan to take the plunge and do my my best to make it work out.

 

[Moneygrubber]

Read the obituaries, not everyone lives to 90 like the expect.

 

[ejman]

I've come to the conclusion that I don't really understand the statement : "At the same time, it is clearly shown in those studies that the returns of the first 10 years practically fully determine whether you'll manage a long financially sound retirement."

I am 63 years old. I ER'd 1/1/2003 = 11 1/2 years ago. Since that time my portfolio's rate of return per Quicken has been 8.45% per year and my NW has increased by 91% since then ( no pension or other sources of income beyond SS which I started at age 62). So, being that I have no idea how long I'll live but it's not inconceivable that I could live another 30 years+ how is it relevant at all what my rate of return for the first 10 years was?.

Realistically, I could be starting retirement tomorrow right where I am at financially and the mother of all Great Economic Depressions be right around the corner. (actually I am starting retirement everyday -what I got is what I got) I just don't see how the prior history in retirement has any bearing on the future.

 

[Totoro]

The descriptions of the axes appear swapped to me.

You are right. Descriptions should be swapped.

 

[Totoro]

I've come to the conclusion that I don't really understand the statement : "At the same time, it is clearly shown in those studies that the returns of the first 10 years practically fully determine whether you'll manage a long financially sound retirement."

I am referring to this article by Pfau: Wade Pfau's Retirement Researcher Blog: Lifetime Sequence of Returns Risk

I am 63 years old. I ER'd 1/1/2003 = 11 1/2 years ago. Since that time my portfolio's rate of return per Quicken has been 8.45% per year and my NW has increased by 91% since then ( no pension or other sources of income beyond SS which I started at age 62).

That's an outcome most of us would sign up for

I just don't see how the prior history in retirement has any bearing on the future.

It does in so far that it affects your starting point today. Since you had a good run-up, your current assets are much higher than they would have been when returns would have been poor. It could have forced you to take out principal, which then is no longer available for a potential later runup.

In addition, you can now withstand a much harder drop in the next period (your withdrawal rate is probably lower now than 10 years ago).

In terms of deciding what to do next you are absolutely right, the past is irrelevant.

 

[Totoro]

Read the obituaries, not everyone lives to 90 like the expect.

Point taken.

I am 34 though, so i've got about a 75% chance to live at least for 40 more years.

Although i'll probably end up in the 25% group if I have to sleep under a bridge

 

[walkinwood]

Research some of the variable withdrawal schemes like Guyton and Bob Clyatt.

You're trying to fund a constant real stream of withdrawals from an uncertain return investment. You can do all the calculations you want, there is not "rule" that will enable you to claim anything approaching certainty.

 

[photoguy]

What's been bothering me about most SWR studies i've seen so far (specifically Bengen & Pfau) is that it is geared towards people with a rather limited future lifespan, with a typical maximum of 30 years.

This means that you can draw from two components derived from your savings:

1. Principal (the balance sheet)
2. Total real return from your portfolio (P&L)

For a (potential) very early retiree this makes those studies incomplete it seems.

Although the data is not presented in the studies, you can use the same simulation methodology (e.g. with FIRECALC) but change the success criteria. I.e., define success as having > X% of the initial portfolio value after 30 years.

Another alternative is to simply extend the number of years you run the simulation in FIRECALC or alternative program.

So why not go for a different approach better suited for long-lived people and very early retirees?

For example: Look at how much *real* return I can expect in the coming 10 years?

I believe Pfau already has a paper incorporating CAPE into SWR studies.

The red arrow is at the current CAPE-10: 25. I did the math and it seems you'll only have a 20% chance of getting at least a real 4% return on stocks starting from valuations above this level

What are your confidence bounds around the 20%? Is it narrow (e.g. 18-22%) or wide (5-50%)?

Also keep in mind that CAPE or PE10 is only slightly more predictive than PE1.

 

[ejman]

I am referring to this article by Pfau: Wade Pfau's Retirement Researcher Blog: Lifetime Sequence of Returns Risk

Thank you Totoro for that link. Very interesting. I also find the follow up comments to that article very illuminating, particularly those of Larry Frank as he articulates concerns that I think are most pertinent.

 

[CaliforniaMan]

...
At the same time, it is clearly shown in those studies that the returns of the first 10 years practically fully determine whether you'll manage a long financially sound retirement.
...

I've come to the conclusion that I don't really understand the statement : "At the same time, it is clearly shown in those studies that the returns of the first 10 years practically fully determine whether you'll manage a long financially sound retirement."
...
Realistically, I could be starting retirement tomorrow right where I am at financially and the mother of all Great Economic Depressions be right around the corner. (actually I am starting retirement everyday -what I got is what I got) I just don't see how the prior history in retirement has any bearing on the future.

I am referring to this article by Pfau: Wade Pfau's Retirement Researcher Blog: Lifetime Sequence of Returns Risk
...

ejman is of course correct, prior history is just that prior history. I think that Wade's chart would be more useful if it were labeled with starting retirement as year 1 instead of year 31. And when we are in withdrawal mode the current year is always year 1. And the next 10 or 15 years has the predomant effect on final success. Of course for very early retirees the chart has less meaning than for later say 30 year retirement.

 

[Lsbcal]

Point taken.

I am 34 though, so i've got about a 75% chance to live at least for 40 more years.

Although i'll probably end up in the 25% group if I have to sleep under a bridge

At 34 ER wasn't even on my radar screen. If you don't have a small fortune, aren't you getting ahead of yourself here?

 

[rodi]

At 34 ER wasn't even on my radar screen. If you don't have a small fortune, aren't you getting ahead of yourself here?

Yes and no... A friend's husband retired at 35. He cashed out when the start up he cofounded was bought. My friend continued working another 15 years to build up her own nest egg (her parents had a BAD divorce and she didn't like risks).

The key - they never expanded their lifestyle. They live in a tiny, paid for, cottage. They don't spend extravagantly. For example - they're foodies - but like to cook the gourmet delights, rather than dine out.

I asked what it was like to have a husband retired so young - she said it was great - he puttered around and did his hobbies... and wasn't the stress ball he was before he retired. She said he was much easier to live with retired, than when he was in start up mode. And he never looked back.

She retired 18 months ago at age 51. So still ER.

 

[Lsbcal]

Nice story Rodi. Sounds like there may have been a small fortune involved.

 

[unclemick]

Research some of the variable withdrawal schemes like Guyton and Bob Clyatt.

You're trying to fund a constant real stream of withdrawals from an uncertain return investment. You can do all the calculations you want, there is not "rule" that will enable you to claim anything approaching certainty.

Hmmm - Layed off age 49, 1993 - lived really cheap in Louisiana swamp - did some temp work - small non cola pension in 1998 - wiped out house wise 2005/Katrina - married a widow in Kansas 2012 with pension/401k.

Agile, mobile and hostile. Agressively varied withdrawal over the years. Especially during 'down' markets and early retirement when I was more 'nervous'. Run calculators several times a year as 'ballpark guideposts' to sense where I'm at - BUT don't always follow them. I like FireCalc and ORP but will sometimes play with others I find.

Post Katrina I find 'stuff' and 'location' much less important and can vary expenses in an extreme manner as/if required.

heh heh heh - It's nice to tell Mr Market to go pound sand when you have learned to be a really cheap SOB. Note that the Woman I married grew up on a farm early on which had no running water, no electricity or central heat, indoor plumbing or silly expensive stuff like that.

Soo if you get your calculations/graph to where you feel comfortable as a ballpark guideline to periodically check while life happens you've got something.

 

[Totoro]

Although the data is not presented in the studies, you can use the same simulation methodology (e.g. with FIRECALC) but change the success criteria. I.e., define success as having > X% of the initial portfolio value after 30 years.

Can you add a target of having 100% in your portfolio after 10 years (100% in real terms, not nominal)? I can't find such an option in firecalc.

Another alternative is to simply extend the number of years you run the simulation in FIRECALC or alternative program.

The issue is that I want to simulate a very long period (50+ years), and that means you'll be using a very limited dataset (1964 is the latest start date!). I think a better proxy is then simulate several 10 year returns and see which ones end up at least as high as your initial portfolio in real returns.

What are your confidence bounds around the 20%? Is it narrow (e.g. 18-22%) or wide (5-50%)?

I didn't simulate, so there is no confidence bound in a statistical sense. I simply took historical 10y real returns with a CAPE above 25 and then counted the ones the returned 4% or more (20% of them, 16 of the 79 periods). The returns vary quite a bit: from -5% p.a. up to 6% p.a. or so. It seems pretty evenly spread between that interval (didn't check in detail yet).

So quite wide, but on average not so good (0% is the average)

I've thought in the mean time though it might be a better approach to look at returns not strictly above a CAPE of 25, but more around a CAPE of 25.

The results look nicer: If you look at CAPEs of say between 22 and 29, you'll have 40% of the periods with a 4% real return or higher (22 out of 58 observed periods). Still not stellar, but less doom & gloom.

And as I said, the analysis is a bit rough right now, if there is appetite I can polish and clean it up a bit and include more recent data

 

[Totoro]

At 34 ER wasn't even on my radar screen. If you don't have a small fortune, aren't you getting ahead of yourself here?

Roughly about $800k saved up. Annual burn rate roughly $30k or so.

In other words, 3.7% post-tax.

Most of it is from a job I quit three years ago, so my earning potential is drastically dropping as time marches on. Been planning for FI since very early on.

 

[Totoro]

Hmmm - Layed off age 49, 1993 - lived really cheap in Louisiana swamp - did some temp work - small non cola pension in 1998 - wiped out house wise 2005/Katrina - married a widow in Kansas 2012 with pension/401k.

That's some roller coaster you got there, and still it ends up ok for you, amazing!

You seem to be a great demonstration of some profound life wisdoms
* Stay flexible
* Marry the right person
* Enjoy the present
* Learn how to live in a swamp

That last one is new for me

 

[NW-Bound]

Swamps also have fauna that help reduce your food budget. I can think of fish, crawdads, alligators, nutria, etc... Some edible flora too, probably, but I do not know much about that.

Out here in the desert, we just cannot live off the land like that (eating cactus?).

 

[Totoro]

Going a bit off-topic, but can you actually eat (not to mention safely catch) an alligator?

How does that work?