USA Today article on SWR

[ERD50]

ERD50: No, the 4% rule has been "The Number" in the broad financial community ever since Bengen's 1994 paper and the Trinity study. ... While it is true that not everybody agreed with it, the broad community did. It's only in the last few years that the general consensus is starting to think that 4% is too high.

OK, I guess I'm just saying that as long as I've been on this forum, there have been a significant group that supports a 3-3.5% WR. It wasn't just in the last few years. I don't know about 'general consensus'. But not a big deal either way.

IMHO, the right SWR to start with when doing initial backtests is what was accepted at the time (notwithstanding that 1972 is 20 years before Bengen's paper). AFAIK, Guyton started from that 4% number when he invented his Withdrawal Decision Rules.

Well, I'm going to apply the method to what I believe is relevant for me. They can use 4% in their paper if they want, that's fine. But my starting point is what has survived 100% of the time in the past. And I'm going to enter my expectations/timing for pension and SS, which can bend these results anyhow.

It's tricky to avoid the hindsight bias fallacy, and IMHO -- for the purposes of comparing these two withdrawal strategies -- using a SWR of 3%-3.5% is hindsight bias.

? FIRECALC uses historical data. That is always going to be in 'hindsight'. It's not 'bias', IMO, it's just using the data you have. Lacking a crystal ball, we can't predict the future anyhow. Just as I would look at historical weather data to plan a vacation, I think it is worthwhile to look at historical data for portfolio survival. Neither are perfect, but watcha gonna do? And the future could be worse than the past, and that is why I like to use a 100% success in my FC runs - why use a plan that has been known to fail in the past?

OTOH, capping the CPI adjustment to 6% is also hindsight bias! The only time inflation has been much above 6% is one period of a few years centered around 1977.

Agree, though I wouldn't call it 'hindsight bias', I'd call it throwing out the data you don't like (cheating!).

So IMHO the proper comparison of plain SWR vs. G-K is 4% and 5.5%, with no inflation caps.

I say use the numbers that make sense for your situation. Why would I use anything but? What would I learn with 4%, if I plan for something else, and have pension/SS kick in later? If you are just looking for a 'generic' idea of whether G-K is 'better' than straight inflation adjusted WR, that could be done for several WRs, to match up to different scenarios (3, 3.5, 4, 4.5, 5....)

As far a putting a "bare minimum" floor, I'm against it. It's all well and good to say, "I insist on taking at least $X", but reality says that when the portfolio gets to zero, you won't be making ANY withdrawals.

IMO, it makes sense as an input parameter to test the 'rule'. I do want to take at least $X - will this, or any other rule, provide for this?

-ERD50

 

[Huston55]

Rayvt - great work on the sheet!

Just one editorial quibble on the prosperity rule wording. Your ssheeet says:

3 PR: If the W/D% computes as less than 20% below SWR, increase the W/D $ by 10%.

It should instead say:

3 PR: If the W/D% computes as more than 20% below SWR, increase the W/D $ by 10%.

 

[imoldernu]

(this is a simple recap of what I have written in other threads, but offered as an alternative to the USA article and the planning that is generally accepted here on ER.)
If we were back in the early days of our retirement, the SWR would have surely been a part of the plan. As it was, the planning was on many dozens of large spreadsheets, using primitve assumptions for inflation, rates of return and actual living costs. As I have previously detailed, our plan works differently, and is based on:

Three different options for expenses... Optimum (highest spending level), Nominal (normal spending), and Austerity (minimum expenses). We adjust these levels as needed depending on circumstances... every year, or whenever a major change is required. These are real calculations, not simply "cutbacks". If necessary, we would sell our Florida mobile home, or our place on the lake. Instead of 2 cars, just one. So far, none of this has been necessary... but the options are there.

Instead of a SWR, we calculate our position by adding:
--- our known annual income from our limited ultra conservative investments... (MM, CD IRA's, I-Bonds, small annuity), and our Social Security.
--- and then taking our total net worth (including housing) and dividing it by the number of years of life expectancy.

This gives us the amount we can spend, and we can adjust accordingly.
Thusfar, after 24 years of retirement, we are only now beginning to tap into the modest net worth.

There are always the "what if's", which we try to cover with nursing home insurance and an emergency fund amount, but the basic plan is built on the expectation that we will someday choose to enter the retirement apartments in our planned senior community, where all expenses...food, transportation and utilities (except phone bills), are covered in a single annual (plannable) fee.

We can remember having a Jumbo 1 yr. CD @ 14%, and projecting inflation at more than 5%, and seeing annuities paying short term 12% with a 5% base, so times have changed.

Not suggesting that what we do/did was right or that it it's even smart... just a different way of looking at retirement that seems to be working for us. Note that the planning is built on a finite life expectancy. ...But if we had based our planning on what current retirement planning charts suggest, we might have spent an extra ten years accumulating necessary funds.

 

[siamond]

I worked a bit more on this terrific Excel sheet from Rayvt, doing things like:
1. adding graphs in constant dollars - much easier to grasp in one look what happens
2. adding more flexible choices of portfolios than 100% S&P500 (thanks to the Bogleheads backtesting data of many investment vehicles)
3. adding the concept of floor and ceiling as 'sanity rules' (easily deactivated if you wish to)

I also would like to find a good way to create upper level graphs similar to Firecalc, showing at once many starting points, but didn't get to that part yet. I also read another article from Mr Klinger with a slightly easier way of defining CPR and PR rules, will give it a try.

I did a lot of testing with my current draft (which incorporates the 3 first items I listed), and this shows really promising results. A more balanced portfolio is notably key to avoid the giant swings in WR that a pure S&P500 model leads to (it actually makes my floor sanity rule nearly pointless).

I'll copy my latest here every now & then...
https://docs.google.com/file/d/0B0svRQGBG_eaTWZjSTFZSERkanc/edit?usp=sharing

 

[rayvt]

Interesting. I used just the S&P because I was mostly interested in seeing the different workings of Guyton-Klinger and standard.

After looking at various scenarios, it struck me that the biggest thing G-K did was to give you absolutely huge draws in the later years. (Assuming the portfolio survived, of course.) I'm not sure just what a 90 year-old is going to do with 3 times the income.
So I incorporated ceiling & floor parameters, which you can set to some multiple of accumulated inflation. It seemed reasonable to set the ceiling at 1.5 and the flooor at 1/2.

New version at same URL:
https://www.dropbox.com/s/cwprtn6y8ouyajj/SPY_Withdraw_by_Guyton_rules.xls
This time, really for sure, I don't expect to make any more changes.

It is really astonishing to see the effect of inflation over a period of 30 years. The cost-of-living goes up 3-4 times. That really shows the pain you'll have if you get an SPIA or IuL without an inflation rider.

 

[Chuckanut]

I
So I incorporated ceiling & floor parameters, which you can set to some multiple of accumulated inflation. It seemed reasonable to set the ceiling at 1.5 and the flooor at 1/2.

New version at same URL:
https://www.dropbox.com/s/cwprtn6y8ouyajj/SPY_Withdraw_by_Guyton_rules.xls
This time, really for sure, I don't expect to make any more changes.

It is really astonishing to see the effect of inflation over a period of 30 years. The cost-of-living goes up 3-4 times. That really shows the pain you'll have if you get an SPIA or IuL without an inflation rider.

Thank you for your fine work on the SS. The article and the SS have opened up some interesting possibilities.

 

[Independent]

As far a putting a "bare minimum" floor, I'm against it. It's all well and good to say, "I insist on taking at least $X", but reality says that when the portfolio gets to zero, you won't be making ANY withdrawals.

The good thing about percentage withdrawal systems is that it's mathematically impossible to completely run out of money.

The bad thing is that the rules don't tell you how bad it can get. If there is no absolute floor, then it's important that people see the minimum withdrawal amounts so they can determine for themselves how often a particular set of rules would have been "successful".

I like the idea of stating that right in the rules. For example, "Taking the greater of 6% of the current balance or an inflated 3% of the beginning balance survived __% of scenarios".

To me, that gets right at the key question.

 

[FinleyG]

All this points to...

It seems clear that there must be flexibility with your WR. So many posters are putting a very low WR in their system, but appear to want to stick with it without variance.

I am disconcerted to note that this appears to be a very bad time to retire if you intend to live off investments. I think that's why there are so many posters intending to live off 2.5% 3% or 3.5%. Problem is, that those percentages are based upon the past (as many have stated).

What is the consensus of the best flexible system. Is it the (G-K?) system in this article, the 95% scenario, or what? The thing that is bothersome, is that you cannot escape W***k unless you have a nest egg that is quite a bit larger than would ordinarily support your lifestyle at 4%.... I guess I have to go to the office this morning.......

Fin

 

[ERD50]

It seems clear that there must be flexibility with your WR. So many posters are putting a very low WR in their system, but appear to want to stick with it without variance. .....

... I think that's why there are so many posters intending to live off 2.5% 3% or 3.5%. Problem is, that those percentages are based upon the past (as many have stated).

Fin

How is this clear? Please show me how some variable WR is clearly better than a fixed low WR.

What if decide that an initial 3% CBP (constant buying power), is the 'floor' of what I want to spend, and 3.5% is my desired level? Is there a variable spending model that will assure I will never go below that 3%? SO far the variable models I've seen take some big dips in spending, and still have scary dips in portfolio (that happens even with zero withdraws).

And what is your alternative to using the past as a guide?

-ERD50

 

[Ready]

I am disconcerted to note that this appears to be a very bad time to retire if you intend to live off investments. I think that's why there are so many posters intending to live off 2.5% 3% or 3.5%. Problem is, that those percentages are based upon the past (as many have stated).

Fin

It strikes me a bit odd that so many people buy into the concept that this is a bad time to invest, yet the stock market is up double digits during the first seven months of this year. I think we are too quick to buy into the media hype telling us what to predict about stock returns over the next decade. The media gets this stuff wrong about as often as it gets it right. I don't believe anyone, even the most brilliant financial experts, can accurately predict what will happen to equities, or even bonds, over the next decade. There is too much uncertainty for anyone to have a crystal ball into the future.

 

[Major Tom]

It seems clear that there must be flexibility with your WR. So many posters are putting a very low WR in their system, but appear to want to stick with it without variance.Fin

My WR is what you might call moderately low, at 2.5%, but I'm doing that not only in the hope that I don't run out of money, but also in the hope that I can increase my WR in the future.

I am disconcerted to note that this appears to be a very bad time to retire if you intend to live off investments. I think that's why there are so many posters intending to live off 2.5% 3% or 3.5%. Problem is, that those percentages are based upon the past (as many have stated).
Fin

I'd love to base my WR strategy on the future, but haven't figured out a way to do that yet (sorry - couldn't resist).

 

[siamond]

I keep doing quite some Excel work and analysis derived from the excellent model Rayvt assembled. Trouble is there are so many parameters to play with that it is hard to extract synthetic information that isn't too specific to a given set of assumptions.

Will share in time, probably the coming week-end.

My preliminary high-level conclusion is that G-K seems clearly superior to the classic model IF you use a diversified enough portfolio, and that a simplified version of G-K is perfectly fine (and easy to enforce). Will provide all sorts of charts to try to back up my assertions...

I also plan to share the outcome on Bogleheads, and brace myself for the tomatoes...

 

[FinleyG]

Ouch!

Well,

Hmm. I agree that if one uses a low enough SWR, it doesn't require a variable formula. Of course, if the funds are invested and the market goes up, you are leaving money on the table. Mind you, that most of my missive was related to the fact that I figure I don't have enough money to retire yet, based upon those parameters. Obviously, if you use a 2.5% SWR, and you leave your money in the bank, you will last 40 years... I think I'm a bit more interested in maximizing (as safely as possible) my retirement.

My concerns about time to retire or the market dangers are just based upon the PE 10, etc... Problem with the market is that it's never totally the same and it reacts to people's fears, so impossible to predict what it will do. We could be on the precipice of a massive runup, or a big correction.... I dunno.

As far as predicting the future, I don't recall doing that.......

That being said, I am definitely a neophyte here and have not even retired yet, just trying to figure it out. I think DW and I need a budget about 75K net of taxes per year and we have no debts plus @2 mil in investments. Looks like we are close, but absent more budget cutting, (and our kids finishing college) we are not ready to retire.

Fin

 

[explanade]

I keep doing quite some Excel work and analysis derived from the excellent model Rayvt assembled. Trouble is there are so many parameters to play with that it is hard to extract synthetic information that isn't too specific to a given set of assumptions.

Will share in time, probably the coming week-end.

My preliminary high-level conclusion is that G-K seems clearly superior to the classic model IF you use a diversified enough portfolio, and that a simplified version of G-K is perfectly fine (and easy to enforce). Will provide all sorts of charts to try to back up my assertions...

I also plan to share the outcome on Bogleheads, and brace myself for the tomatoes...

What is the "classic model" the fixed WR or 3 to 4%?

Does the spreadsheets generate the withdrawals and balances under the G-K and the "classic model" for comparison?

 

[ERD50]

If you thought the previous comments were worthy of an 'ouch!' response, better get your first-aid kit out!

... Obviously, if you use a 2.5% SWR, and you leave your money in the bank, you will last 40 years...

Not only is it NOT 'obvious', it is wrong (historically). You forgot about inflation. Do some FIRECalc runs, and you will see that "money in the bank" will not support a buying power adjusted 2.5% WR for 40 years, or even 30.

As far as predicting the future, I don't recall doing that.......

Then what is your alternative to historical data?

Of course, if the funds are invested and the market goes up, you are leaving money on the table.

I am sooooo less worried about this, than I am of running out of money in my later years. And I can, and will, recalculate as I age.

That being said, I am definitely a neophyte here and have not even retired yet, just trying to figure it out. I think DW and I need a budget about 75K net of taxes per year and we have no debts plus @2 mil in investments. Looks like we are close, but absent more budget cutting, (and our kids finishing college) we are not ready to retire.

Fin

Sounds like you are doing well (I didn't catch your age). Try some FIRECalc runs, enter any SS/pensions you expect to get a better handle of how this would have done in the past. Ask questions, a lot of posters here will try to help.

-ERD50

 

[siamond]

What is the "classic model" the fixed WR or 3 to 4%?

Does the spreadsheets generate the withdrawals and balances under the G-K and the "classic model" for comparison?

Classic model is X% (I typically look at 4%, but it's a parameter) of initial portfolio, then inflation-based adjustment. The usual Bengen idea.

And yes, there is a full comparison between the various models, including charts of withdrawals and portfolio value.

Give me a few more days, every time I think more about it or run some scenario, I figure out a small thing or another to tweak, or I draw some add'l conclusion or question a premature conclusion... But I think I'm getting somewhere.

 

[urn2bfree]

The doom and gloom types are basing their negativity on current predictions of bond and stock returns -some based on PE10. The problem with this negativity is that

1)PE10 concerns mostly the US stock market and therefore may miss the effect of a more diversified portfolio including REITs and Emerging markets, etc
2) while the bond predictions for low yields for a while may be somewhat reliable, the PE10 is a weak predictor. From a very interesting piece about the power of portfolio diversity in enhancing SWR- SEE THIS ARTICLE http://advisorperspectives.com/newsletters13/The_Power_of_Diversification.php which includes the following observation:
"The r-squared of the linear relationship between current yield and future 10-year annualized return for the Ibbotson Intermediate-Term Bond index was 92%. The r-squared between the PE10 and future 10-year annualized return was 24% (r-squared is a standard measure of the strength of a relationship; higher r-squared corresponds to higher predictability)"

This tells me that the predictions of below average returns even in the less diversified S&P 500 made according to the PE10 are more likely wrong than right.

 

[Independent]

It seems clear that there must be flexibility with your WR. So many posters are putting a very low WR in their system, but appear to want to stick with it without variance.

I am disconcerted to note that this appears to be a very bad time to retire if you intend to live off investments. I think that's why there are so many posters intending to live off 2.5% 3% or 3.5%. Problem is, that those percentages are based upon the past (as many have stated).

What is the consensus of the best flexible system? Is it the (G-K?) system in this article, the 95% scenario, or what? The thing that is bothersome, is that you cannot escape W***k unless you have a nest egg that is quite a bit larger than would ordinarily support your lifestyle at 4%.... I guess I have to go to the office this morning.......

Fin

There is no consensus.

For starters, there's no consensus on how to compare different withdrawal patterns. Rule A gives more money in early years in some scenarios, Rule B gives more money in the later years in those same scenarios. And, when you look at other scenarios, that relationship is reversed, etc. How do I determine whether A or B is "better"?

Siamond says he is working on comparisons. I'll be interested to see if that generates any discussion.

 

[rayvt]

Y'know when I said that I wouldn't be making any more changes. I lied.

Per comments by various people, I figured it would be useful to test with other asset allocations besides 100% S&P500. So this latest update also includes data for 1-yr and 10-tr Treasuries. Just enter the relative weights of the allocations.

Same URL:
https://www.dropbox.com/s/cwprtn6y8ouyajj/SPY_Withdraw_by_Guyton_rules.xls

========================================================
ERD50:

Please show me how some variable WR is clearly better than a fixed low WR.

There is no possible answer, because it all depends on what each persons prefers. De gustibus non disputandum est.

It's all a balancing act between getting enough withdraw money, maintaining (or growing) income to match inflation, maximum acceptable drawdown, terminal value, and not running out of money in your lifetime. You can't have everything you want.

=====================
What was encouraging to me was that with a 60/40 allocation that the lowest portfolio of both G-K and standard 4% strategy was essentially the same -- about $2000 apart. For a really bad time to start, just before the 1973-74 bear market. Indeed standard had higher withdraws for the first 20 years, but at the cost of lower portfolio value.

Setting the start to 1965 was a similar story, the standard had larger draws for about 25 years. But then the standard 4% portfolio crashed and hit $0 in 2008, but the G-K portfolio was $473K.

There is no "right" method. But what is clear is that there are times when you must reduce your withdrawals if you want the portfolio to survive. The thing I find attractive about the G-K method is that the rules for changing your withdrawals are defined and not seat-of-the-pants.

 

[ERD50]

....
There is no possible answer, because it all depends on what each persons prefers. De gustibus non disputandum est.

It's all a balancing act between getting enough withdraw money, maintaining (or growing) income to match inflation, maximum acceptable drawdown, terminal value, and not running out of money in your lifetime. You can't have everything you want. ...

Agreed, that was really my point. There's only so much you can do and keep withdrawals within some usable range. I'd be very surprised that a particular rule would be 'clearly' better under a wide range of time periods.

If a portfolio takes a 40-50% hit in buying power, and your WR was in the range of 4%, I don't think there is a lot of leverage there. But it will be interesting to try some things out, to see how much of an effect the adjustments can make over time. I'd imagine you could pull a few more years out of the portfolio by easing up on WR while you are down.

-ERD50

 

[Chuckanut]

There is no "right" method. But what is clear is that there are times when you must reduce your withdrawals if you want the portfolio to survive. The thing I find attractive about the G-K method is that the rules for changing your withdrawals are defined and not seat-of-the-pants.

+1

My biggest concern is that when I am in my very old age, will I still be able to do the calculations and make the adjustments? Heck, will I even remember that I need to to them? At some point, I wonder if it wouldn't be better to just dump it all into the Pssstt Wellesly fund and trust them to continue to work their magic. One could do worse.

 

[scrinch]

FinleyG, I'm with you. It's clear to me that a flexible WR is better than working and saving up so long that I can put my retirement withdrawals on fixed, inflation - adjusted, 100% failsafe autopilot. Maybe it's not better (or clear that it's better) to others though. It's really a personal choice how much risk each person is comfortable with taking, and how to manage that risk in retirement.

As far as what investment performance to base a model on, in my view it shouldn't be the past and it can't be the future.... it should be based on your view of the future. You shouldn't just use past data because you don't have anything else to use. If you think past performance is the best indicator of future performance, then go ahead and use it. But if you think investment performance in a 7 billion inhabitant, globalized economy, post - industrial world will be fundamentally different from those in a post-WW2, cold war, flower child world, then you should do your best to build those differences into your model.

 

[Huston55]

First, I want to say that this is a great thread; worthy of being placed in the "Best Of" for easy reference.

I've read Clyatt's book and tested his method with FireCalc, and read the G-K papers which include data on the number of 'cuts' and 'freezes' that various WD rates would required; that data seems very consistent with the modeling results from Rayvt and Siamond.

I plan to use a variable withdrawal rate versus a fixed one because it seems to fit my situation best, and that's what I'd like to see some discussion on because I expect my situation is not uncommon.

All of this discussion on WD methodology ignores other income sources (pensions, SS or even annuities), which will make a substantial difference. While not everyone has a pension, most (maybe all) will receive SS. When these other income streams kick in, the required WD, whether fixed or variable, will/can go down. I know we can model this with FC, Fido RIP or other tools, and I do. But, the 4% fixed SWR and G-K WD rules don't take into account those other income sources, which will ameliorate the drops in any variable WD approach. This is one of the key factors in my decision to use a variable WD methodology because, the planned timing of my retirement and these other income sources make 'the worst case' not so bad for us.

I'd like to hear how other are accounting for this in their plans.

 

[rayvt]

Other income sources don't really come into play when you are looking at the topic of SWR.

SWR is the topic of taking withdrawals from retirement account(s) -- de-cumulation. How much income you need is a completely different topic.

What you are probably thinking about is changing your withdrawal at different points in time. Like taking a very high (unsustainable) rate now with the expectation of reducing it to a very low rate in the future when another income stream kicks in.

This is very personal to each person's individual circumstances, so it'd be rather clumsy to add this to the usual tools, like FC, etc.

What you could do with a spreadsheet is to put in your (high) SWR and current account value, and then note the value in X years (when the step comes in) and then re-enter that value and the new (low) SWR, and then check the final outcome.

If you want to make the effort, you could enhance the spreadsheet to have an initial SWR and then use a different SWR at a different date.

 

[Katsmeow]

This is very personal to each person's individual circumstances, so it'd be rather clumsy to add this to the usual tools, like FC, etc.

I don't think it is clumsy to add it to the usual tools. More work for sure, but not clumsy. There is a manual spending mode on Firecalc and the Fidelity Plan all you to vary spending.

For typical retirees who aren't retiring until they all of their income streams are online (SS, pensions primarily) then these calculators that assume a level WD rate are useful. But, for early retirees where it is common to withdraw more early on, then those type of calculators are not particularly useful.

I'd like to hear how other are accounting for this in their plans.

I have a huge difference in WD between now (I'm not on SS yet and our expenses are higher than in future due to kids still at home) and later. So I mostly have used Firecalc and Fidelity RIP. When I've occasionally used others I have projected where I think our Portfolio will be at the time I will take SS and then model withdrawals from that point. I've tried to use varying models both optimistic and pessimistic.