30 Year Withdrawal Rates - Fun with numbers!

[Dtail]

What's the inflation rate?

Let's say 3% nominal return with a 2% inflation rate.

 

[NW-Bound]

Andre Tobias once wrote "a luxury once sampled becomes a necessity." Hence, none of my spending is discretionary. I have to have all of this stuff.

That has been one of my favorite quotes too.

Now, have you learned to stop sampling more luxuries?

 

[NW-Bound]

At the risk of starting the first back-and-forth ER forum tirade of 2019, I do think this is where a SPIA can come into play by bringing some known stability as a foundation.

Variability over the long haul will exist for sure, but one can contain it to some extent by creating a base level of certainty. Obviously, there are return trade-offs that come along with this approach or similar approaches that rely more heavily on very safe fixed income securities.

OK everyone, let's be nice now...

I don't think people here are against SPIA. It may work for some.

For me, the SS to be claimed, the dividend from stocks, and the cash I hold form that base level of certainty.

 

[Scratchy]

Very nice, thanks for sharing your analysis!

I'd be very curious how much factoring in an expense ratio of around 0.07 would change the withdrawal rates.

 

[Nature Lover]

At the risk of starting the first back-and-forth ER forum tirade of 2019, I do think this is where a SPIA can come into play by bringing some known stability as a foundation.

True statement, and no need for a back-and-forth tirade. Long discussions all over the web have sought to define/debate how to best address the need to accommodate life's variability, and many retirement financial strategies (some of which include SPIAs) are designed to incorporate both a more stable portion and a more variable portion of planned annual budgets/funding/WD's.

There are many roads to Dublin.....we all choose the strategy(s) that works best for us given our own situation.

 

[Chuckanut]

Exactly! --- That is why you should Plan on variability... Withdrawal methods such as VPW, do exactly that. You cannot control variability, but you can react properly to it.

+1

Thanks for reminding me I need to calculate my withdrawal for 2019.

 

[Cut-Throat]

+1

Thanks for reminding me I need to calculate my withdrawal for 2019.

I just did mine this morning....

 

[FREE866]

Interesting thread.

I run two differnt sets of numbers--one for 18 years as that is when I hit age 70. Then one for 25 years assuming I die at 95.

Would really like to be able to wait until 70 to collect social security, but if we have a horrendous market before then that could change.

 

[NW-Bound]

If I may ask - what is your discretionary expense as a % of total budgeted expenses?

Not the guy you asked, but we are planning on about 40% discretionary.

Me neither, but we're about 75% Discretionary..... Living in Hawaii this winter....

It’s been around half (after taxes), maybe a bit more these days,

Andre Tobias once wrote "a luxury once sampled becomes a necessity." Hence, none of my spending is discretionary. I have to have all of this stuff.

That has been one of my favorite quotes too.

Now, have you learned to stop sampling more luxuries?

I looked at my Quicken screen, and asked myself if push comes to shove, do I need this expense, or that expense?

I then added them all up, and they came to 50% of what I spent. This means my WR could be down to below 1.5%.

At that point, there's no more travel, no more gifting, nor donation, etc... I would still have my 2nd home, but no money to fix anything if the roof leaks, or the stain fades, or a car needs replacement, etc...

Life would be dull, but not miserable. I still eat the same food, drink the same booze. Could be a lot worse...

 

[audreyh1]

+1

Thanks for reminding me I need to calculate my withdrawal for 2019.

+3 I did mine today too. Between 12/24 and 12/31 I got quite a raise!

Down from last year though.....

 

[big-papa]

I've been off work for the holidays, the weather has mostly been bad, so I've been playing with spreadsheets! Here's what I did:

Using historical returns for S&P 500, 1 yr T-Bill, 10 yr Bonds, and annual inflation rates I calculated the maximum inflation adjusted withdrawal rate that would result in zero remaining savings after 30 years, starting each year between 1928 and 1988. These are sequential calculations for each 30 year period, they are not randomly picked historical returns and inflation rates.

I used an asset allocation of 60% stocks, 10% T-Bills, and 30% Bonds.

The results are shown in the attached plot.

Some Statistics:

• The minimum successful withdrawal rate was exactly 4% (1966)
• The average successful withdrawal rate was 6.4% with a standard deviation of 1.7%

I'm interested in your thoughts and observations. Has anyone seen a similar analysis elsewhere?

Take a look at the Simba Spreadsheet, available over on Bogleheads. It does what you did, and more. It has a large number of asset choices and it's maintained. A new update with 2018 results will be added in a few weeks.
The overview (which also contains a link to the post that contains a link to the spreadsheet itself. https://www.bogleheads.org/wiki/Simba's_backtesting_spreadsheet


I find it to be a fantastic resource and it's pretty easy to work with. It doesn't do variable withdrawals (the spreadsheet is big enough as it is). Like others have mentioned, a withdrawal method whose withdrawal has some connection to recent returns makes more sense to many of us. You are trading more certainty that your money will last as long as you need it to with the fact that the amount you withdraw will vary year upon year. VPW (also available on Bogleheads) and its variants, fixed percentage withdrawal, and Guyton-Klinger are among the many choices out there that do this. There are posts on this forum about it as well as over on Bogleheads.

Cheers
Big-Papa

 

[marko]

Yes, and FIRECalc goes more than 50 years farther back, starting in 1871.

Just a question not a challenge (and a bit off topic) :

Maybe I don't understand the benefit of going back so far.

I wonder how relevant applying data from 1871 is in today's world.
Doesn't that possibly skew results one way or another? The 1800's were another planet economically.

Wouldn't using data from modern economic times, say, 1970 (or even 1930) forward give a more accurate picture?

 

[audreyh1]

Just a question not a challenge (and a bit off topic) :

Maybe I don't understand the benefit of going back so far.

I wonder how relevant applying data from 1871 is in today's world.
Doesn't that possibly skew results one way or another? The 1800's were another planet economically.

Wouldn't using data from modern economic times, say, 1970 (or even 1930) forward give a more accurate picture?

Personally I’m glad those rougher earlier time periods are included in Firecalc. It increases my confidence in the model. After all we’re looking for worst case scenarios. For my scenario the toughest start years are pre 1926 and 1966 only comes in 12th worst or something like that.

 

[ERD50]

..... For my scenario the toughest start years are pre 1926 and 1966 only comes in 12th worst or something like that.

Can you explain that? I've always seen 1966 popping up as the worst start year in the database.

Does your scenario start off with low withdrawals in the early years or something?

-ERD50

 

[conversationalphrase]

Take a look at the Simba Spreadsheet, available over on Bogleheads. It does what you did, and more. It has a large number of asset choices and it's maintained. A new update with 2018 results will be added in a few weeks.

The overview (which also contains a link to the post that contains a link to the spreadsheet itself. https://www.bogleheads.org/wiki/Simba's_backtesting_spreadsheet





I find it to be a fantastic resource and it's pretty easy to work with. It doesn't do variable withdrawals (the spreadsheet is big enough as it is). Like others have mentioned, a withdrawal method whose withdrawal has some connection to recent returns makes more sense to many of us. You are trading more certainty that your money will last as long as you need it to with the fact that the amount you withdraw will vary year upon year. VPW (also available on Bogleheads) and its variants, fixed percentage withdrawal, and Guyton-Klinger are among the many choices out there that do this. There are posts on this forum about it as well as over on Bogleheads.



Cheers

Big-Papa

Thanks! I will check it out.

Happy New Year.

 

[big-papa]

Personally I’m glad those rougher earlier time periods are included in Firecalc. It increases my confidence in the model. After all we’re looking for worst case scenarios. For my scenario the toughest start years are pre 1926 and 1966 only comes in 12th worst or something like that.

Our view of the future using the past will always have uncertainty to it. Not every scenario that could happen in the future has already happened in the past. There are many tools to try and get a better view including
- going back in time as far as possible with a backtest. It is good to be aware of the sources of some of the really old data, including the uncertainty involved in recreating indices that didn't actually exist at the time as well as corporate governance rules that may (or may not) have existed at the time.
- using Monte Carlo methods (of which there are many variants) which may/may not include bootstrapping and autocorrelation.

Those are main two methods of which I'm aware but at the end of the day, investing still takes a leap of faith since none of us have a crystal ball. Recalling the lessons of "significant digits" back in science class, it's probably best not to apply a lot of precision when using a dataset that isn't always precise.

 

[big-papa]

At the risk of starting the first back-and-forth ER forum tirade of 2019, I do think this is where a SPIA can come into play by bringing some known stability as a foundation.

Variability over the long haul will exist for sure, but one can contain it to some extent by creating a base level of certainty. Obviously, there are return trade-offs that come along with this approach or similar approaches that rely more heavily on very safe fixed income securities.

OK everyone, let's be nice now...

Can't speak for everyone, but on this forum and others, SPIAs generally aren't considered a bad thing - except in the context of low interest rates. The flame-wars generally start when one starts discussing other types of annuities that have high, hidden costs.

I have a TIPs/Ibonds ladder that I'll be using starting at age 70 as a supplement to SS. At age 85, depending on whether I'm still on this earth, my health and portfolio status, it probably makes a lot of sense to consider a SPIA to help carry me over the finish line.... As always YMMV.

 

[corn18]

I am always considering getting a 15 year period certain SPIA @ age 55 as a bridge to SS. I price one every once in a while to see what the equivalent interest rate is. Right now it sits @ 1.7% APR, which is less than 5 year CD's and less than a Vanguard MM fund, so I let it go. There are no mortality credits @ 55 (and no COLA) so it seems a better approach is a bond tent to get me to SS.

 

[big-papa]

I am always considering getting a 15 year period certain SPIA @ age 55 as a bridge to SS. I price one every once in a while to see what the equivalent interest rate is. Right now it sits @ 1.7% APR, which is less than 5 year CD's and less than a Vanguard MM fund, so I let it go. There are no mortality credits @ 55 (and no COLA) so it seems a better approach is a bond tent to get me to SS.

Agree - with low interest rates and not much in the way or mortality credits, I wouldn't consider it either. There are SPIAs (not many) with inflation protection, but there is a pretty steep cost for it in that nominal payments today would be significantly lower than just a regular SPIA.

There are many ways to bridge to SS. Some do as you suggest and use 5 year CD's. Some use a TIPs ladder in order to have a guarantee of inflation protection. Still others use Ibonds for the same reason, though you have to plan farther in advance for it because of the limitation on how much you can purchase each year.

Finally, what I've seen some do is to perform an NPV on all future cash flows (like SS) to calculate a virtual portfolio and calculate the withdrawal from your portfolio based on that calculation. No change to the portfolio itself and nothing set aside specifically for the bridge. It does mean you're withdrawing more from your portfolio until SS kicks in, so if that bridge is long and there's a bad sequence of returns, it could put a strain on your portfolio. But I would argue that even if you set money aside for a bridge, it's still part of your portfolio - whether you count it or not.

 

[audreyh1]

Can you explain that? I've always seen 1966 popping up as the worst start year in the database.

Does your scenario start off with low withdrawals in the early years or something?

-ERD50

I don’t think it is, although it’s often referenced. The default Firecalc 4% setting has 5 failure years. According to another thread

the 5 worst failures were 1907,1929, 1937, 1965,1966.

it doesn’t give the order of which fails the fastest.

I was running %remaining portfolio models with 50/50 total stock market and 5-year treasuries. Various withdrawal rates and hunting for worst year - largest portfolio drawdown - during a 30 year period.

 

[big-papa]

I don’t think it is, although it’s often referenced.

I was running %remaining portfolio models with 50/50 total stock market and 5-year treasuries. Various withdrawal rates and hunting for worst year - largest portfolio drawdown - during a 30 year period.

Using Simba's spreadsheet from Bogleheads with 50% Total Stock Market and 50% Intermediate Treasuries yields the following for 30 year retirements using data starting in 1871 with 1966 still coming up as the worst year to have retired when using a fixed real % withdrawal (meaning this is the starting percentage for year 1 and is increased by the previous year's inflation rate each year)

1899: 4.2%
1906: 4.0%
1965: 4.0%
1966: 3.8%
1968: 3.9%
1969: 3.9%

All other years are higher than these.

Cheers,
Big-Papa

 

[big-papa]

Using Simba's spreadsheet from Bogleheads with 50% Total Stock Market and 50% Intermediate Treasuries yields the following for 30 year retirements using data starting in 1871 with 1966 still coming up as the worst year to have retired when using a fixed real % withdrawal (meaning this is the starting percentage for year 1 and is increased by the previous year's inflation rate each year)

1899: 4.2%
1906: 4.0%
1965: 4.0%
1966: 3.8%
1968: 3.9%
1969: 3.9%

All other years are higher than these.

Cheers,
Big-Papa

The above was the SWR (meaning maximum withdrawal that results in $0 at the end of 30 years). There's another calculation the Simba Spreadsheet can do which is PWR (Perpetual Withdrawal Rate) which is defined as the maximum which can be withdrawn each year (inflation adjusted, just like SWR) such that on the last year you have the full starting value of your portfolio.

for PWR, it looks like this:
1890: 2.3%
1892: 2.3%
1902: 2.3%
1965: 2.5%

All other starting years are higher than these.

Cheers,
Big-Papa

 

[REWahoo]

Kitces has this on his website, showing the 30 year SWR for a 60/40 AA beginning in 1871.

1937 and 1966 look very similar to me.

Also of interest, 87% of the time you would have been fine with a SWR of 5%.

 

[big-papa]

Kitces has this on his website, showing the 30 year SWR for a 60/40 AA beginning in 1871.

1937 and 1966 look very similar to me.

Also of interest, 87% of the time you would have been fine with a SWR of 5%.

It's somewhat AA dependent. Switching what I did above with Simba to a 60/40 portfolio I see the following for SWR.

1937 at 4.6%
1966 at 3.9%
1968 at 3.9%
1969 at 3.9%


1966,68 and 69 are the lowest in this model.


Cheers,
Big-Papa

 

[FREE866]

Kitces has this on his website, showing the 30 year SWR for a 60/40 AA beginning in 1871.

1937 and 1966 look very similar to me.

Also of interest, 87% of the time you would have been fine with a SWR of 5%.

I could spend hours on that site! haha
what is the SWR for higher equity percentages though?