SWR in bad times

[sgeeeee]

I've been running simulations again. I was wondering if you optimized your stock allocation for specific, very bad single retirement dates in the past, would some pattern emerge. To do this study, I first ran SWR v6.1 with the following assumptions:
(1) January values for the S&P500
(2) 50% of the annual withdrawal taken at the beginning of the year, 50% at year end.
(3) inflation indexed to the CPI.
(4) fixed income series based on TIPS@2%
(4) 0.20% annual expense ratio
(5) portfolio rebalanced annually

Next I considered two allocation cases:
(I) I used a 60% stock allocation and increased the initial withdrawal rate till I got 10 failures (ie 10 starting years that resulted in portfolio bankruptsy).

(II) I used a 30% stock allocation and increased the initial withdrawal rate till I got 10 failures.

Between the two cases above, I came up with 13 unique years that produced failures with one or both stock allocations.

Finally, I simulated retirement starting in January of each of the failure years and altered the stock allocation in increments of 10% to identify the optimum allocation for retirees who retired during each of these tough times.

Here's the results:

............Optimum ....................
............Stock ....................
............Allocation...................
YEAR....(stock/TIPS).......SWR
1893......100..................5.52%
1903........90..................4.87%
1907........60..................4.63%
1910........70..................4.52%
1929........40..................4.66%
1930........60..................4.66%
1962........70..................4.64%
1964........60..................4.46%
1965........30..................4.32%
1966........10..................4.34%
1967........50..................4.53%
1968........30..................4.46%
1969........30..................4.53%

The optimum allocation if you consider all years is 30% and the overall SWR is 4.30%.

I was surprised at what little this effort showed me. There are a couple of thing that might be worth noting. For one, the initial retirement years that were bad because of high inflation tended to exhibit lower optimum stock allocations than other bad years. A second result that surprised me is that only one year (1966) produced an optimum stock allocation that was lower than the overall optimum allocation. In reality, the optimum allocation may have been shifted downward for three other years, but if that shift were less than 10%, it wouldn't have showed up in this study.

Hi SG,

Quote:

The optimum allocation if you consider all years is 30% and the overall SWR is 4.30%

I'm am not sure at how you arrived at the 30% for all years and the 4.3%.

Here is what I have been playing lately

I have also wondered (and have even asked Dory) if FIRECalc would consider uneven Withdrawal rates. In other words, take into account normal human behavior. Such as the Year following the market, where losses exceeded a certain percentage(say 10%) to reduce the SWR by a certain percentage Say 20%. And in the year following a market gain of another percentage (say 10%) to increase the SWR by a certain percentage say 20%.

It would be interesting to see what the SWR would be in these cases. My guess is that it would be a lot higher by avoiding drawing down an already down portfolio. This actually mirrors human behavior, in that major purchases (a new car) would be done after a banner year in the market.

[TheFIREman]

What you want is already there:

http://home.golden.net/~pjponzo/sens...ithdrawals.htm

Improve your income in good times, cut back in the bad.

[unclemick]

Does anybody else waste time thinking why the SWR always seems to come within hand grenade distance of the historical dividend yield? Maybe the recent 20 yr or so downtrend in div. yield is a blip and we're due for RTM.

FIREman,

Thanks for the great link. It may take me the rest of the winter to digest this stuff though.

The formulas are downright scary!

Pg1g2g3...gN = fPi1g2g3g4...gN + fPi1i2g3g4...gN + fPi1i2i3g4...gN + ... + fPi1i2i3...iN

[TheFIREman]

Yeah. Made me wish I stayed awake in my 1st year algebra classes.

Fireman,

I am trying to play with this spreadsheet and some of the red boxes do not seem to labled and I am unsure of what to put in them. An example is the two boxes at the top next to Small Cap/Value. The have percentages in them but I'm not sure what they relate to?

Do you know ?

Thanks

[wzd]

And if you go to http://home.golden.net/~pjponzo/gummy_stuff.htm you can keep yourself busy for several winters. I didn't have the link at the time, but this is the page that has information on tax impact of withdrawals - and fights conventional wisdom that deferral is always better.

Wayne

[sgeeeee]

Quote:

Hi SG, I'm am not sure at how you arrived at the 30% for all years and the 4.3%.

That's the answer that intercst's SWR v6.1 gives you when you optimize your allocation for a 30 year draw using TIPS @ 2% and using all the other assumptions I listed. If you do the same optimization on FIRECALC you will get a slightly different answer, but it is very close.

[TheFIREman]

Quote:

Fireman, I am trying to play with this spreadsheet and some of the red boxes do not seem to labled and I am unsure of what to put in them. An example is the two boxes at the top next to Small Cap/Value. The have percentages in them but I'm not sure what they relate to? Do you know ? Thanks

Click on the Explain-3 sheet to find out. They basically give you a range so that you can see which percentage of returns are less than the top number and which are greater than the bottom. I like to try scenarios where I put 15% in the bottom and see it come out more than 50% of the time. Now that's return.

I have customised this sheet to allow for MERs, sliding allocation (equity = 100 - age) and for the implication of tax (Canadian resident capital gains). It is surprising how little tax impacts the overall returns. If you want to suggest something to Peter (gummy), go right ahead, he is very ameniable to making changes. What a nice guy.

[Michael]

unclemick wrote:

Does anybody else waste time thinking why the SWR always seems to come within hand grenade distance of the historical dividend yield? Maybe the recent 20 yr or so downtrend in div. yield is a blip and we're due for RTM.

Mike replies:

It intuitively makes sense. Even a 90% drop in the S&P will not deplete your portfolio if you live on the dividends, and don't touch the principal. The cash buffer will tide you over during the times when dividend yield falls temporarily, as over the long term dividends have tended to rise to keep up with inflation. There are no long term studies at current dividend yields because they have never been this low before. If you could figure out the reason for their fall, you might be able to guess at whether or not they will rise again. My uneducated guess is that it may have something to do with a combination of the IRA/401k tax laws and the unusual size of the boomer cohort. Of course, since there have never been 401k/large cohort combinations in US history, there is no way for me to reasonably figure out what the likely outcome will be.

[earlyout]

SWR,

One of the worst times to retire was from 1966 to 1995 (30 year period). Over this time, the first 17 years the real return on stocks was zero. The next 13 years were very good. Had one retired (in 1966) all SWR above 4% and with any combination of stocks/bonds,etc ended up broke before the 30 years were up (based upon inflation adjusted withdrawals).

For those inclined, this time period could be used as a test of any portfolio or SWR they want to consider. Cut-throat has a very valid consideration in adjusting withdrawal based upon actual bear/bull performance of your portfolio.

For more info see limk:

http://www.efficientfrontier.com/ef/998/hell.htm

Hi earlyout! Now there is some excellent support for my
"no stocks" position. A 17 year period with zero
return from stocks. I NEVER assume that I have anywhere near 17 years to work with (I would feel this way at any ER age by the way) If it happened
before it could happen again. Nuff said!

John Galt

[Hiss]

Quote:

One of the worst times to retire was from 1966 to 1995 (30 year period). Over this time, the first 17 years the real return on stocks was zero.

That is interesting. I wonder now how FIREcalc uses the PPI or CPI selected. Does anyone know if it simply takes an average over all the years of the market, or does it look at the specific PPI or CPI for those particular years of the market for which it is analyzing to come up with the success prbability?

[Michael]

Quote:

Does anyone know if it simply takes an average over all the years of the market, or does it look at the specific PPI or CPI for those particular years of the market for which it is analyzing to come up with the success prbability?

It is my understanding that it uses the individual year's CPI to adjust that year's withdrawal, not an average number. It does a very good job of determining what would have worked during past markets, back to 1871. As long as the future is not worse than the past, the SWR numbers will work in the future. Be aware that there is no way for any of us mortals to know whether the future will be worse than the past or not.

Quote:

Now there is some excellent support for my "no stocks" position. A 17 year period with zero return from stocks.

Not Really John Galt, because a zero return during the high inflationary periods of the 70's was pretty good! At least money was not lost due to inflation!

I am almost counting on this happening again during my retirement. And that is exactly why I want to have stock postion of 50% with a well diversified portfoilo.

If by chance I should make it to 80 years old. My stock position then should only be 20%.

Cut-throat........I see your point (re. zero return on stocks/inflation/etc). However, using my system
you just sail along cashing your interest checks
and ignore all the "stock" nonsense. I confess
that I did own a few stocks in my 40s and later.
Now though, nearly 60 and 100% retired, I'm sorry
I just can't see it.

John Galt

Fireman,

Thanks for the info. This looks like a very good tool.

I have been studying it and cannot seem to find how you add a bond component and cash component to the Asset Class Mix?

Can you help with this

[TheFIREman]

For cash you will have to roll your own. It is quite easy to do so by adding another column to the end of the asset classes (column BJ). I did that for pseudo-TIPS by taking the inflation rate and adding 2% (a bit optomistic, I know). Fixed income is already covered by t-bills, 5 year and long bonds. The numbers circled in red 1 2 3 4 are the asset class selectors and they start numbering at 1 at column AZ. So AZ is 1 is large cap growth, BA is 2 is Large cap value,.... T-bills is BD is 5... and so on. When you change a number in red, you will notice that the label in the top left-hand corner changes to the corresponding asset class. Take a look at that end of the sheet and you should have no probs

[unclemick]

Looks to me (being math challenged) that gummy has brought the power of the spreadsheet and some good math to 'compensate for the SD effect in such a way as to create an 'annuity' type line - which allows you to spend more early and take the principle toward the end of span.

Should work? One caution - don't use an asset class that deviates from it's historic growth trend for 20 yrs. or so. The selected portfolio will still depend on (historic?) growth rates for the asset classes selected.