Equalizing your personal WR% for comparison

[ERD50]

A common problem here is that discussing your personal WR% isn't as simple as it sounds. Often, an ER will have a high WR% in their early years, which declines when pensions/SS kick in, and then may increase again if that pension is non-COLA. So what number should we use when trying to convey the relative safety/risk of our overall or average WR?

Here is my suggestion:

A) First, you must disclose your time period and success rate assumptions, as these are key variables (I'll think about this later, but maybe everything could be referenced to the default 30yr/95% success...).

B) In FIRECalc, enter your time period, and $1,000,000 portfolio without any pensions or SS entries. Ignore spending amount.

C) Use the investigate tab, set:

Given a success rate, determine spending level for a set portfolio, or portfolio for a set spending level

Search for settings that will get a success rate of as close to __XXX___ % as possible (usually within 1%) by changing... _check__ Spending Level​

and SUBMIT.

D) Take the resulting spending level, and shift the decimal place to get the WR%.

Example:

45 years, 100% success results in a $32,393 spend level, so that is 3.24% WR. So any combination of pension and portfolio and spending that provides 100% success for 45 years could be expressed as a 3.24% effective WR%. So that WR% should represent an equalized value for comparison.


As a simple validation of this, compare:

A) A retiree with a $50,000 spend level who can take $25,000 SS from the start. The other $25,000 (50%) comes from a 4% withdraw from a $625,000 portfolio.

B) A retiree with a $50,000 spend level with no pension/SS. The $50,000 comes from a 4% withdraw from a $1,250,000 portfolio (2x the size of retiree A).

So they both are drawing 4% from the portfolio, and the portfolios will have the same success rate. Of course, if they fail, retiree A still has half their spending sourced by SS, but that is separate from the portfolio success itself.​

If your plan shows money left over at the end of your time period (>100% success), you could either simply say 'lower than x.xx% WR', or adjust spending in FIRECalc to get a similar ratio of end/begin portfolio, and report that WR%.

Make sense?

-ERD50

 

[bradaz2488]

I have also been looking at my WR% throughout my ER plan. I'm 50 and ER'ed last May. My ER plan is also for 45 years. I use MS Money for my main ER planning and FIRECalc says I have 100% success rate. In MS Money I'm able to track my medical expenses separately and apply a higher inflation rate of 7.5% vs. the 2.5% for all other expenses. Not able to do this with FIRECalc. This really increases the % allocated just for medical to my total annual expenses which in turn increases my WR% each year to coverall all our expenses for that year. Below is a table should the % allocated the medical expenses and the respective WR% for my ER plan:

Age %Medical WR%
50 6% 2.4%
60 13% 3.3% Note: DW will retire when I'm 57.
70 17% 4.1%
80 24% 5.0%
90 37% 6.7%

Not sure if this is realistic or not or if I should be concern about how I have allocated medical expenses in my plan....

 

[galeno]

Bill Bergen of the 4% rule has a nice table regarding SWRs. My country has a good socialized healthcare system which we use for the big stuff e.g. cancer. We also buy private insurance for fast access and better service for the little stuff.

Age % Stocks SWR
55 65 4.2
60 60 4.3
65 55 4.4
70 50 4.5
75 45 5.0
80 40 5.2
85 35 5.5
90 30 6.0

 

[Lsbcal]

Some good thoughts in this ERD50. Is this for a constant spending model only i.e. inflation adjusted spending? It would seem to be logically so.

I might add one more step between C & D:
Select a value for "Leave some money in the portfolio for my estate". This is not just for estate purposes but to define a minimum portfolio (floor) value, e.g. for a 1,000,000 portfolio we might not want the value to ever dip below 300,000.

 

[ERD50]

Some good thoughts in this ERD50. Is this for a constant spending model only i.e. inflation adjusted spending? It would seem to be logically so.

I might add one more step between C & D:
Select a value for "Leave some money in the portfolio for my estate". This is not just for estate purposes but to define a minimum portfolio (floor) value, e.g. for a 1,000,000 portfolio we might not want the value to ever dip below 300,000.

Yes, I assume spending is adjusted by inflation each year, like the FC defaults.

I tried to cover your case in my last paragraph of the OP. So if you want some 'dip protection', your WR will be lower, and you would need to mimic that 'dip protection' in the FC 'investigate' run you make to get the 'effective WR'.


-ERD50

 

[Midpack]

Here's my projected WR % by source by year http://www.early-retirement.org/forums/f28/the-shortcomings-of-swr-after-retirement-68147.html, and some other members...

 

[Lsbcal]

Yes, I assume spending is adjusted by inflation each year, like the FC defaults.

I tried to cover your case in my last paragraph of the OP. So if you want some 'dip protection', your WR will be lower, and you would need to mimic that 'dip protection' in the FC 'investigate' run you make to get the 'effective WR'.


-ERD50

Just a caution, I've kind of rediscovered a nasty bug in FC that has the summary reporting much worst then actual minimum value portfolio results. This is for a constant spending model where you can get at the spreadsheet. I now only trust FC's spreadsheet results. My guess is this bug is well known but it's amazing that it has not been fixed.

I'm not entirely sure about the "Given a success rate, determine spending level ..." results. One cannot get a spreadsheet out to check the results. I definitely would not trust setting the min portfolio value at all times value. If running I'd recommend just set it to zero. Then put the spending that you get into another run and check the spreadsheet results.

 

[ERD50]

Just a caution, I've kind of rediscovered a nasty bug in FC that has the summary reporting much worst then actual minimum value portfolio results. This is for a constant spending model where you can get at the spreadsheet. I now only trust FC's spreadsheet results. My guess is this bug is well known but it's amazing that it has not been fixed.

I'm not entirely sure about the "Given a success rate, determine spending level ..." results. One cannot get a spreadsheet out to check the results. I definitely would not trust setting the min portfolio value at all times value. If running I'd recommend just set it to zero. Then put the spending that you get into another run and check the spreadsheet results.

Thanks, I kinda forgot I started this thread!

I think your double-check with the spreadsheet is a good idea. I'm not sure I'm aware of the bug you refer to. I have not done many runs using that minimum setting - I do think it is valuable, but I have a fuzzy recollection that the results maybe were not clear. Maybe I wasn't convinced that it really meant that the interim values never dropped below X, or just the ending value?

One way I checked this was to shorten the years to 2, 3, 5, 7, 10 etc, and see if any ending portfolios were below my specified minimum.

I recall being confused by the spreadsheet output though - are you sure this is a bug, or is it a matter of inflation versus non-inflation number adjustments? I seem to recall that the spreadsheets use one method, while the display is different?

-ERD50

 

[daylatedollarshort]

The Fidelity RIP has a detailed cash flow analysis which shows a year by year withdrawal rate based on the parameters input. Our WR is all over the map due to inflation and changing life circumstances - kids, college, SS in different years, non-cola pensions, downsizing, ACA to Medicare costs, etc.

 

[Chuckanut]

Thanks for posting your method of using this FireCalc feature. Since the Detroit pension ruling I have been rerunning FireCalc with various reductions in both pension payments and a reduced or eliminated COLA. And, of course, different withdrawal rates to compensate.

 

[Lsbcal]

...
I recall being confused by the spreadsheet output though - are you sure this is a bug, or is it a matter of inflation versus non-inflation number adjustments? I seem to recall that the spreadsheets use one method, while the display is different?

-ERD50

[-]My check on the spreadsheet was pretty careful. I looked at the calculation example spreadsheet to make sure I understood how they got the cells in the full sequences spreadsheet. The cells in the full sequence spreadsheet should be inflation adjusted just as in the summary page.

I then ran a standard simulation that shows all the pretty wiggly lines. FIRECalc reported the minimum and maximum values for the all simulations in its summary. That minimum number was way off according to that run's own spreadsheet.

I could come up with an example run if someone wants to see this or check me for accuracy. I have been known to make mistakes in life, believe it or not . I kind of trust the spreadsheet results after having check a detailed calculation. Anyway, user beware![/-]

EDIT: I took another look and sure enough the spreadsheet results are not inflation adjusted -- thought I'd covered that one. So to convert the spreadsheet to the summary results, one would need the matrix of inflation factors. Perhaps doable, I'll have to think about this tomorrow. I don't know why it was done this way as it's inconvenient for my purposes.

Anyway, sorry about the false alarm. FIRECalc is probably just fine.

 

[Lsbcal]

See my mea culpa above.