SWR rule-of-thumb assuming principal preservation

[Delawaredave]

Been reading posts and done some Firecalc simulations.

Undersand that 3-4% SWR provides a "90-ish" percent chance of a "40-ish" year withdrawl term - allowing for depletion of original capital.

With the same assumptions, what would be the SWR assuming one wanted to preserve purchasing power of initial principal ?

Example: say you have $1million - how much can you withdraw over 40-ish years to have $1million real dollars (todays purchasing power) at end to pass to kids.

Thanks ! Delawaredave

 

[Cut-Throat]

A tough question to answer because you have to make some assumptions.

So here we go. Let's say inflation averages 3% - You get a total return of 5% over the life of the portfolio. You could withdraw 2% per year and still maintain Principle.

 

[retire@40]

A tough question to answer because you have to make some assumptions.

So here we go. Let's say inflation averages 3% - Yout a real return of 5% over the life of the portfolio. You could withdraw 2% per year and still maintain Principle.

I believe you mean nominal return, not real.

 

[Nords]

Example:  say you have $1million - how much can you withdraw over 40-ish years to have $1million real dollars (todays purchasing power) at end to pass to kids.

You can have whatever's left over after inflation & taxes.

From "Triumph of the Optimists" I'm learning that inflation has run at about 3.5% over the last 100+ years, but about 5% over the last 30 years. So pick your personal inflation number from there.

The Gordon Equation (an intimidating mathematical concept that attempts to predict future returns) comes up with numbers between 5-7% for a portfolio split between, IIRC, Treasury bonds and the S&P500. (Someone leap in and correct me here before I flounder much deeper.) The 10% number quoted for "stocks since 1926" has several assumptions, inaccuracies, & omissions that render it great for past performance comparisons but useless for prediction.

Your taxes (and expenses like trading or portfolio "management") will vary but will probably chew up 0.5-1.5% of the remainder. Unless you're using a financial advisor, in which case their fees will ensure that there isn't any remainder.

You can see that even an optimist will have your SWR down into the 2-3% range and a pessimist will just send you back to the workforce.

Your salvation lies in the realm of "new" international investing (with fewer investing barriers and more countries growing faster than the U.S.), low-expense ETFs & funds, a wider variety of non-traditional investments like commodities & REITs, the growth (or the return of) dividend investing, and hopefully a continued small-cap value premium.

I know that at this point someone will say "Why not just put it all in TIPs or I bonds and harvest whatever's left over after inflation"? That should work too, as long as CPI approximates your personal inflation rate (but cat food is not a good "hedonic adjustment") and as long as you can get 4% over CPI.

Please let me know if you find an inflation-adjusted investment paying 4% over CPI...

 

[sgeeeee]

Hi DD,

You can figure this out with FIRECalc.  Here's a run I just did when I saw your question:

Withdrawals: set to $40K but let FIRECalc find the real answer.
Starting Portfolio: set to $1M
Lifespan of Portfolio: 30
Withdrawal change 1: set to $1M starting in year 30
percent portfolio in stocks: 50
select: CPI
left remainder of items as default.
select: summary results

The answer you get is that the 4% rate defined by your inputs is about 56.1% successful.  A 95% success rate requires a 3.07% initial withdrawal rate.

Obviously you can customize this to your own situation and needs.

By the way, notice that the way I ran the simulation, I left the inflation adjustment tab for withdrawal 1 checked.  So I was requiring that I preserved the initial portfolio inflation adjusted.  If you only want to preserve the nest egg without inflation adjustment you can uncheck the box.

The non-inflation adjusted answer:  at the 4% rate defined by your inputs you are about 90.2% successful.  A 95% success rate requires a 3.82% initial withdrawal rate.

One detail of this technique is that you only require that the nest egg be preserved in the final year of life (as specified by your Lifespan of Portfolio input.  It does not require that the nest egg remains always above the initial value.  It could slump in year 10, for example and recover.   

 

[Cut-Throat]

I believe you mean nominal return, not real.

Yes R40, I corrected my post!

You can see that even an optimist will have your SWR down into the 2-3% range and a pessimist will just send you back to the workforce.

Nords,

Naw ! - Thats Hoc*s talk! - Just plan on drawing down Principle gradually. Let the kids fend for themselves!

 

[Nords]

Naw ! - Thats Hoc*s talk! - Just plan on drawing down Principle gradually. Let the kids fend for themselves!

No "kidding"-- I'm not the one trying to preserve principal for my progeny!

 

[Delawaredave]

Hi DD,

You can figure this out with FIRECalc.  Here's a run I just did when I saw your question:

Withdrawals: set to $40K but let FIRECalc find the real answer.
Starting Portfolio: set to $1M
Lifespan of Portfolio: 30
Withdrawal change 1: set to $1M starting in year 30
percent portfolio in stocks: 50
select: CPI
left remainder of items as default.
select: summary results

The answer you get is that the 4% rate defined by your inputs is about 56.1% successful.  A 95% success rate requires a 3.07% initial withdrawal rate.

Thanks for your simulations - I ran below and got different answers.

Simulation 1:   $1mm, 40 year period, all other defaults -- a $42,200 annual withdrawl survives 95% of the time (7 "red" periods, 2 additional ones below $1mm)

Simulation 2:  Same as above, but dropped withdrawl to $40,000.   There were only 5 periods where final number fell below $1mm (so this is at least 95% of the time).

So:
- a 4.2% withdrawl achieves 95% positive ending portfolios. 
- a 4.0% withdrawl achieves 95% ending portfolios above $1mm

Does this mean that 0.2% reduction preserves the orginal principal ?

I'm suprised the reduction is that small - appreciate any thoughts

 

[sgeeeee]

Dave,

I see a few problems with what you are trying to do. 

Not accounting for inflation:  In my earlier simulations I included a withdrawal change in the final year of the simulation that was an inflation adjusted $1M.  I used the withdrawal change option and selected the inflation adjusted box for that withdrawal.  What that does is subtract an inflation adjusted $1M from the portfolio in the final year.  If the portfolio drops below zero, you didn't preserve the initial portfolio.  If it doesn't, you did.  So the success rate you get when you do that is exactly the success rate you are interested in.  That's the only way to know that an inflation adjusted $1M is what you are comparing to.

When you look at the final column, you are seeing an amount that is not inflation adjusted.

40 years is too long:  One of the problems with historical simulators is that the number of unique simulations is limited by history.  When you select a 40 year simulation, the most recent full sequence started in 1962.  You miss one of the worst times in history to have started retirement that happens just a few years later.  If you are interested in retirement times longer than about 30 years, its probably a better idea to run the simulations for 26, 28, 30, 32 years, and then draw a plot and look at the assymptotic response.  At 40 years, the ommission of recent start dates is beginning to skew the data and produce results that are slightly optimistic.

When I run your 40 year case ($1M portfolio and all defalts) then include the preservation of an inflation adjusted $1M withdrawal in the final year,  I get a 3.80% SWR.  This compares with 4.22% SWR without capital preservation.

The defalt 75% stock allocation is important to the superior results.

Here's what you find for your 40 year, all defalt case

. .STOCK . . . . . . . . . SWR w/
ALLOCATION . . . . PRESERVATION OF PRINCIPAL
0% . . . . . . . . . . . . . 0.00%
25% . . . . . . . . . . . . 1.04%
50% . . . . . . . . . . . . 2.75%
75% . . . . . . . . . . . . 3.80%
100% . . . . . . . . . . . 3.90%

 

[Delawaredave]

Thanks a lot ! I didn't realize "final column" numbers weren't inflation adjusted.

Agree with your comments about 40 years.

Overall, looks like a 1/2% withdrawl rate reduction to preserve purchasing power of original portfolio. Seems right.

Have a great day !

 

[Spanky]

Been reading posts and done some Firecalc simulations.

Undersand that 3-4% SWR provides a "90-ish" percent chance of a "40-ish" year withdrawl term - allowing for depletion of original capital.

With the same assumptions, what would be the SWR assuming one wanted to preserve purchasing power of initial principal ?

Example: say you have $1million - how much can you withdraw over 40-ish years to have $1million real dollars (todays purchasing power) at end to pass to kids.

Thanks ! Delawaredave

There is also another calculator from T Rowe Price:
http://www3.troweprice.com/ric/RIC/

The Best SWR is only 2.9% (or $2,900 per month) for a portfolio of $1 MIL, retirement length of 40 years and simulation success rate of 90%.

If the retirement length reduces to 35 years, the SWR becomes 3.1%.

The best success rate is a portfolio with 40% stock, 40% bond and 20% short-term securities while the worst is 100% equity?

Try the calculator yourself.

 

[sgeeeee]

The T. Rowe Price calculator uses a monte carlo approach so it does not have the same limitations for long retirement periods that the historical simulators like FIRECalc do.

Monte carlo simulators have their own limitations and tend to provide slightly pessemistic results. I also don't see a way to get the T. Rowe Price calculator to require initial portfolio preservation.

 

[mb]

SG, what is the basis for saying that Monte Carlo simulations produce slightly pessismistic results. Is that based on comparisons with history-based simulations like FIRECALC or is there something else in addition to that?

MB

 

[sgeeeee]

SG, what is the basis for saying that Monte Carlo simulations produce slightly pessismistic results.  Is that based on comparisons with history-based simulations like FIRECALC or is there something else in addition to that?

MB

Hi MB,

First, monte carlo simulators do not have to produce pessimistic results. Since the programer can decide on the return distributions and inflation distributions the program will use, any values could be chosen -- producing as optimistic or pessimistic results as desired.

But generally, what monte carlo programmers do is to use historical distributions of the S&P 500 returns, historical distributions of bond returns, historical distributions of inflation, . . . for whatever assets they want to consider. Monte carlo programs then produce random numbers to select the returns and inflation consistent with the historical distributions. It will use a complete set of these results for each year of the simulation, compute the implied performance, then pick a new set and compute performance for the next year. At the end of the simulation, the performance distribution for each asset and inflation is consistent with the historical record.

This methodology neglects correlations between the returns and inflation and neglects correlations in performance due to recent history. We know, for example, that inflation affects stock and bond performance. While we can't write an exact formula for the relationship, a relationship exists. Countless other correlations exist which cannot be captured mathematically. These correlations tend to limit the range of values for various returns. Neglecting correlations will tend to provide results that are worse than reality.