motivated by the discussion of the SWR Research Group at NFB
I've been looking in on the discussion of Safe Withdrawal Rates on NFB and ...
>SWR? Again?! You've done that
before!
Yes, I know, but it's quite interesting. You know that I've always felt it was a crazy scheme to withdraw, say 4% of your original portfolio
(increasing with inflation), ignoring current market machinations.
In fact, my favourite scheme is to modify this year's withdrawal rate according to how much (or how little) you made last year.
>That's sensible withdrawals, right?
Yes. Last year's gains would determine how much you withdraw.
However, in that NFB discussion forum, they're suggesting that the criterion be based upon the current value of
E10/P.
>Huh?
That's average Earnings for the past 10 years divided by the current stock Price ... a variation of E/P.
For example, a portfolio consisting of 80% stocks + 20% bonds would have had an annual
E10/P something like Figure 1.
>Something like?
Well ... uh, the numbers are taken from Schiller's site.
They apply just to the S&P 500, but we'll use them for our 80% stocks + 20% bonds portfolio. In fact, we'll use Schiller's E10/P data for
every portfolio, regardless of the components of the portfolio. Those E10/P numbers will serve as "market indicators" and ...
>Is that legal?
The proof is in ...
>Yeah, yeah. The proof is in the pudding. But why the red line which ends in 1965?
Aah, good question!
One of the worst periods for investments in the past umpteen years started in 1966.
Has you withdrawn the notorious 4%, increasing with inflation, your porfolio would have died in about 27 years.
Using last year's E10/P value to determine your withdrawal for the current year would have done better.
 Figure 1

The comparison was like this:
Figure 2
See the red guy? That's what would have happened to a $100K initial portfolio had you withdrawn
the notorious 4%.
>Increasing with inflation?
Yes. By 1993 it was dead.
Now suppose we had withdrawn at a rate which was the notorious 4% rate ... but modified to reflect the amount
by which E10/P (for the previous year) differs from the average.
Note: The "average" we're talking about is the cumulative average from 1928 and is shown as a thin
magenta line in Figure 1.
In fact, the green curve is your $100K portfolio had your withdrawal rate been calculated
each year as follows:
 You calculate the notorious 4% rate (which increases with inflation). We'll call it X%
This changes every year, of course.
 You calculate the difference between last year's E10/P and the cumulative average of E10/P values up to the previous year, since 1928
(or some convenient time in the past). For 1966, that average would have been the 7.1%, as indicated in Figure 1.
We'll call this difference D%
... and this should be positive after a "good year".
 You then withdraw a percentage of your initial portfolio equal to:
0.90 (X% + D%)
>What's that 0.90 out front?
You scale down, withdrawing just 90% of X% + D%. (The 90% is some kind of "safety factor")
But note that, after a "good year" (when E10/P exceeds the average, since 1928)
the difference D% is positive, so you'd withdraw MORE than X%.
>And after a bad year?
Then D% would be negative so you'd withdraw less than the notorious 4% rate.
See 1973? That wasn't a good year! (See the magenta dot in Figure 1.)
With the standard, garden variety 4% rule you'd have withdrawn about 6.5% in 1974.
That's 4% increased by inflation. Remember! In this the "standard withdrawal ritual"
you ignore the market and your portfolio gains and losses and just withdraw a percentage of your starting portfolio, increased by inflation.
If your starting portfolio was $100K and you're withdrawing 4% (increasing with inflation), then, by 1974, you'd be withdrawing 6.5% of $100K
or $6500.
But, in 1973, E10/P was less than the average to 1973.
That gives you a heads up, so you'd withdraw less in 1974 (namely 4.5%).
In fact we'd have:
 The "usual withdrawal ritual" would give a 6.5% withdrawal for 1974 ... so X% = 6.5%
 The average E10/P from 1928 to 1973 is 6.8% ... see the blue dot in Figure 1
 The 1973 value of E10/P is 5.3% ... see the magenta dot in Figure 1
 We calculate D% = 5.3%  6.8% = 1.5% ... reflecting the fact that last year weren't that good
 In 1974 we withdraw 0.9 (X% + D% ) = 0.9 (6.5%  1.5%) = 4.5% ... as in Figure 3
 Figure 3

The spreadsheet (such as it is) looks like so:
There's an "Explain Sheet" which looks like this: Click!
To download the spreadsheet RIGHTclick here and Save Link Target
... or, sometimes clicking here will work !
P.S.
The spreadsheet may change without notice.
>Isn't that what everybuddy says?
I guess. But remember:
 Schiller's E10/P values used by the spreadsheet are January values of the
moving average of earnings over the previous 120 months ... that's 10 years.
 That gives equal weight to the month 10 years ago as it does to the current month.
 If you wanted to weight the current month earnings more heavily, you could use the
Exponential Moving Average
... that's EMA
 The spreadsheet gives you the choice: Schiller's E10/P moving average or the exponential guy
... which we call E(10ma)/P
