GARCH

Research on Safe Withdrawal Rates

Moderator: hocus2004

Post Reply
User avatar
gummy
*** Veteran
Posts: 107
Joined: Wed May 14, 2003 4:00 am
Location: Hogsville, ON
Contact:

Post by gummy »

P.S.
Rather than going thru years of SWR research on NFB, I think I'll wait till you guys finish your tutorial

... then I'll understand everything :lol:
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

Rather than going thru years of SWR research on NFB, I think I'll wait till you guys finish your tutorial

... then I'll understand everything


Oh sure. But the tutorial is probably going to leave out all of those ever most edifying intercst rants. It just won't be the same!
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

gummy wrote:Maybe that's why I prefer "last year's market performance" rather than a valuation involving a ten year average.
You got my interest because this is a new measure of valuation for me. It has been quite a while since I tried something new. [The last time that I did, I made all sorts of variations based upon Tobin's Q and mixed it in with everything that I could think of as well. P/E10 still came out best.]

I have made graphs with HDBR50 and HDBR80 Historical Surviving Withdrawal Rates from 1921-1980 versus:
1) The total real return from the previous year.
2) The total real return starting from two years earlier.
3) The total real return for the same year. [Technically, this is an illegal predictor. It assumes knowledge of something that happens after the prediction is made.]

None of them worked out.

The R-squared values were 0.0121 and 0.0166 from the previous year. The R-squared values were 0.035 and 0.0466 starting from two years earlier. The R-squared value of HDBR80 was 0.0017 when using the total return of the same year.

Remember that Historical Surviving Withdrawal Rates are influenced greatly throughout the first decade. Professor Shiller showed that P/E10 works well in predicting the total return ten years later, which is the intermediate term. Using the market's behavior from the previous year or two years strikes me as a short-term predictor.

Have fun.

John R.

BTW, the New Tool makes predictions based upon the market's total return (without deposits and withdrawals) over several years. The shortest time that was useful was around 6 years. Ten years did a good job. Fourteen years was the best. R-squared exceeded 90%.
User avatar
gummy
*** Veteran
Posts: 107
Joined: Wed May 14, 2003 4:00 am
Location: Hogsville, ON
Contact:

Post by gummy »

The thing about HDBR (or HSWR or HFWR or ...) is that they are constant withdrawal rates, fixed for thirty years.
The only thing that changes the amount withdrawn is inflation.
If inflation were zero, you'd withdraw a constant dollar amount each year.
I'm not sure what the correlation between some constant HDBR and total returns the previous year tells us.

My suggestion is to change the withdrawal rate every year, based upon current portfolio performance.
If you made a bundle in 2004 on your particular portfolio, then why would you not remove some for that trip to Grand Cayman Island?

We can argue as to what constitutes a "bundle", but whatever definition we provide, it means a gain much more than you need to pay the bills.
Further, any argument which looks at historical "successful" withdrawal rates, based upon gross market performance, is of little use to the investor who puts all his money in company stock ... like my son-in-law Confused

What I like about "sensible withdrawals" is that it's devoted to your particular portfolio ... and it changes to suit your current portfolio performance.

Anyway, I'll wait for the next instalment of that tutorial :D
unclemick
*** Veteran
Posts: 231
Joined: Sat Jun 12, 2004 4:00 am
Location: LA till Katrina, now MO

Post by unclemick »

I guess I'm quasi -gummy: portfolio yield plus lagniappe. All those various SWR calc.'s are there as guide posts so I stay on the playing field.

Dumb question - does GARCH or er the calcs posted here pick up major inflection points? Media pontification/speculation goes on and on about the awaited rise in interest rates -presumably to affect SWR in future.

Does a trip down history lane show trends or sudden breaks in SWR? I'm thinking Central Bank vs commercial paper in the early days, The Great Depression, the 60's go-go years, 73-74 break, the interest trend since 1982 and so on.

My impression is: a relatively sudden break and then a trend is established which goes on for a while.
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

"My suggestion is to change the withdrawal rate every year, based upon current portfolio performance."

For Gummy and others: I am certainly open to suggestions.

I suggest that we build upon these results:
The 4% Shocker dated Tue Jan 04, 2005.
http://nofeeboards.com/boards/viewtopic.php?t=3238

These findings were based upon withdrawing 4% of a portfolio's current balance. There are no separate adjustments for inflation. When the portfolio does well, withdrawals increase. When the portfolio does poorly, withdrawals decrease.

What I have reported are 5-year moving averages of withdrawal amounts in real dollars. This smooths out the year-to-year fluctuations.

I am able to look at any combination of 1) removing a constant percentage of the portfolio's initial balance (plus inflation), 2) removing a constant percentage of a portfolio's current balance (regardless of inflation) and 3) removing a fraction of the gains, but none of the losses, from the previous year. OTOH, I would have considerable difficulty putting in an upper limit to withdrawals, which is part of your Sensible Withdrawal approach.

Have fun.

John R.
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

unclemick wrote:Does a trip down history lane show trends or sudden breaks in SWR? I'm thinking Central Bank vs commercial paper in the early days, The Great Depression, the 60's go-go years, 73-74 break, the interest trend since 1982 and so on.
No. Changes in Historical Surviving Withdrawal Rates have been gradual. They can be grouped into three dangerous eras: 1) between the turn of the previous century and World War 1 (IIRC), 2) the Great Depression and 3) the 1960s-mid1970s (associated with stagflation).

You can see this for yourself using FIRECalc. The visual effect is impressive.
http://capn-bill.com/fire/

I recommend that you look at portfolios with 50% stocks and with 80% stocks with a withdrawal rate of 5% and a 30-year lifespan. At 5%, the withdrawal rate is large enough for there to be a reasonable number of failures, but not too many.

Generally speaking, the 50% stock results are very smooth. With 80% stocks, there is a fair amount of choppiness, especially associated with the Great Depression.

Have fun.

John R.
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

Here is another contribution to Gummy's tutorial. This gets added to the end of the first section (just before the discussion about statistics).

Starting at 4:
If the equation is (for example) :
y = 0.490x + 0.029
and the most recent
E10/P = 0.035 (or 3.5%)
then we get
y = 0.49*0.035 + .029 = 0.046 or 4.6% as the Calculated Rate for the coming year. This is our best estimate of what history will record as the Historical Surviving Withdrawal Rate when looking back in 2035.
5. To calculate the Safe Withdrawal Rate, we must identify the lower confidence limit. (We use a 90% confidence level.)
6. A quick and easy way to approximate the 90% confidence interval from our existing plot is to remove some of the data points that are farthest away from the regression line.
7. The line has 61 data points corresponding to the 61 years in the interval of 1920-1980. Ten percent of this is 6.1. Rounded, this is 6. We remove the 6 data points farthest from the regression line.
8. Using our eyeball to make our choices, we remove the five data points above the line with Historical Surviving Withdrawal Rates of 9% and greater. We remove the single data point below the line (for the year 1921) which has the highest earnings yield and a Historical Surviving Withdrawal Rate close to 9.8%.
9. The data that remains range between +1.4% or +1.5% above the regression line and -1.8% below the regression line. The total range of the data above and below the regression line is 3.2% or 3.3%.
10. Our approximate confidence interval is plus and minus 1.6% to 1.7% (for a 90% level of confidence).
11. We choose to use the bigger number, 1.7%.
12. The Safe Withdrawal Rate equals the Calculated Rate minus 1.7%. The High Risk Rate equals the Calculated Rate plus 1.7%.
13. In our example, the January 2005 Calculated Rate is 4.6%. The January 2005 Safe Withdrawal Rate is 2.9%. The January 2005 High Risk Rate is 6.3%.
14. The amount one actually withdraws is his Personal Withdrawal Rate. Most people would choose something slightly above the Safe Withdrawal Rate of 2.9% but well below the Calculated Rate of 4.6%. Some might limit withdrawals to less than 2.9% based upon their assessment of the future, especially for the stock market. Very few people knowledgeable of these findings would withdraw 6% or more.

Have fun.

John R.
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

Back on page 1, gummy wrote:
Two are the sound (or Audio: left and right sounds for stereo) and three are the Video: the Red, Green and Blue (RGB) components of the scene ... presumably with colour filters.
The RGB Video is then reconstructed to generate three related Video signals:

the brightness (or Luminance, called Y = 0.30R + 0.59G + 0.11B)
two colour signals: (U or CR = R - Y and V or CB = B - Y, for example)
These rearranged Chrominance signals may be labeled CR and CB or U and V or I and Q or Hue and Saturation.
Note that, given Y, U and V, one can recover R, G and B by solving three equations in three unknowns.
>My TV knows how to
I remember learning about color TV in my High School days.

The reason for transmitting the Luminance signal was so that color programs would work with the existing black and white sets.

The color processing was quite sophisticated, especially for the times. (Analog circuits using vacuum tubes.)

The color signals were transmitted in phase quadrature centered around the color carrier frequency (3.58 MHz?). The carrier was suppressed throughout the picture, but a gated sample of the carrier was sent between (horizontal?) sweeps. A special circuit, which used a piezoelectric crystal for ringing (or in a phase locked oscillator), would recover the color carrier.

In any event, the two color signals were chosen specifically to make it easier to match (Caucasian) skin tones. These were the most important. Pictures with green or blue people don't look realistic.

The color signals and the phase of the carrier and the rest of the circuits eventually led to some precision resistors, known as the color matrix. The color matrix converted the three signals (the luminance and the two color signals) into red, green and blue.

Gummy's "TV knows how to" do this. It has the equivalent of the color matrix.

RCA made a revolutionary discovery.

They could eliminate the expensive, high precision color matrix circuit by changing the phase of the color carrier. In essence, an inexpensive, small, adjustable capacitor replaced a boat load of nonstandard, precision components. [The actual circuitry requires a little bit more than I have indicated, but not much.]

Nobody thought about RCA's simple, eloquent solution until color TV had been around for years.

Have fun.

John R.
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

gummy wrote:Here's something to think about:
...
5: Then we win the 2006 Nobel prize :D
P.S.
I have a scheme much simpler than GARCH that does a similar thing.
Mine is much like the riskmetrics model.
Not only will your approach be simpler, it will also be better.

When people look at something like GARCH, they think that it must be accurate.

When people look at your simpler model, they will not assume that your model is accurate. They will be careful.

The reality is likely to be that your model will be at least as accurate as GARCH and much, much better. People will believe what GARCH produces because of its credentials. They will forget to think. They will make expensive mistakes. With your simpler model, they will keep their eyes open. They will be much better off. They will remember to protect their wallets.

Have fun.

John R.
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

hocus2004 wrote:My understanding (a greatly limited one) of the Sensible Withdrawal Rate strategy is that it calls for smaller take-outs following bad return years and larger take-outs following good return years. The benefit to this is that you are not stuck going with a number derived from looking at the worst-case-scenario returns sequence. By agreeing up front to limit your take-out in some circumstances, you are able to increase your take-out in other circumstances. I see a good bit of appeal in this idea since in all likelihood you are not actually going to personally experience a worst-case returns sequence.
In The 4% Shocker, withdrawals equaled 4% of the portfolio's current balance. There were no caps on the withdrawal amount. Without such caps, the (5-year average) amount withdrawn at year 20 fell to as far down as 2.8% of the initial balance (in terms of real dollars). [This was for a sequence in the mid-1960s.]

When making Sensible Withdrawals, Gummy limits the maximum amount that he withdraws in any year.

This is a very important distinction.

Have fun.

John R.
User avatar
gummy
*** Veteran
Posts: 107
Joined: Wed May 14, 2003 4:00 am
Location: Hogsville, ON
Contact:

Post by gummy »

Back on page 1, gummy wrote ...

... then comes stuff about colour TV.
John Where's this page 1 you speak of?
(The 50th birthday of colour TV was last year
... and the first sets cost about as much as a car :D )

Anyway, in my colour TV tutorial (on my website)**, I mentioned that our TV would arrive shortly.

It arrived yesterday and it took me a whole day to figure out how to connect our satellite receiver, VCR, DVD and distribute signals throughout the house. :?
(The Missus has a small TV in the kitchen.)
http://www.gummy-stuff.org/TV-stuff.gif

** The very first tutorial I ever wrote was in 1957 and it was on (would-you-believe) colour TV!
I remember the horizontal sync signals sitting on the "back porch" of the horizontal sweeps and the vertical sync in the 45 lines of black space at the end of the vertical sweep.

A few years ago I had a lousy TV which didn't respond to the vertical sync and I'd see all the digital data go by (from my satellite provider) at the end of the vertical sweep :D
User avatar
gummy
*** Veteran
Posts: 107
Joined: Wed May 14, 2003 4:00 am
Location: Hogsville, ON
Contact:

Post by gummy »

Not only will your approach be simpler, it will also be better.
I'm always reluctant to say "This is better than that."
My usual stance is to say "I think this is better than that".

I did a google search of my website for "I think":

http://www.google.com/custom?q=%22I+thi ... -stuff.org

You'd think I did a lot of thinking :lol:

Anyway, your latest addition to YOUR tutorial has now been incorporated.
User avatar
gummy
*** Veteran
Posts: 107
Joined: Wed May 14, 2003 4:00 am
Location: Hogsville, ON
Contact:

Post by gummy »

P.S.
For pedagogical reasons (when your tutorial is complete), I think I should colour the various definitions (HSWR, HFWR, Calculated Rate, Safe Withdrawal Rate, High Risk Rate, etc.) so the reader will recognize them when they occur

... as I did here (near the bottom):
http://www.gummy-stuff.org/parabolas2.htm
User avatar
gummy
*** Veteran
Posts: 107
Joined: Wed May 14, 2003 4:00 am
Location: Hogsville, ON
Contact:

Post by gummy »

another P.S.

Since I was interested in the relationship between a year's E10/P and that years' Portfolio Return (for various portfolios) I made up a spreadsheet:

http://www.gummy-stuff.org/SS/E10-P.xls

I think it's quite interesting:




Another thing that I find interesting (as noted in the spreadsheet):

The correlation between a year's Portfolio Return and the next year's Return is often negative and that means that, after a good year, you might consider removing some "extra" money for that trip to the Bahamas ... else some of the gains may get annihilated the following year ;)
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

I'm always reluctant to say "This is better than that."
My usual stance is to say "I think this is better than that".


May your kind multiply in days to come, Gummy.
unclemick
*** Veteran
Posts: 231
Joined: Sat Jun 12, 2004 4:00 am
Location: LA till Katrina, now MO

Post by unclemick »

Second That

You guys are on a roll. Keep up the good work.
User avatar
gummy
*** Veteran
Posts: 107
Joined: Wed May 14, 2003 4:00 am
Location: Hogsville, ON
Contact:

Post by gummy »

Hey hocus:
Concerning "I think...",
a wise man once said:
Don't forget, gummy, this stuff may someday stand the world of conventional investing advice on its head. If that happens ...
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

gummy wrote:Back on page 1, gummy wrote ...

... then comes stuff about colour TV.

John Where's this page 1 you speak of?
(The 50th birthday of colour TV was last year
... and the first sets cost about as much as a car :D )
Page 1 is page 1 of this thread. Scroll to the beginning or end of a set of posts and you will the the page number. Look at the upper left hand corner and the lower left and right hand corners.

The table of contents has page numbers as well.

Have fun.

John R.
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

More about gummy's color TV tutorial:

I notice that CR and CB are the color outputs when using RCA's method of demodulation.

These matrix equations are very easy to solve.
L = R + G + B
CR = L - R
CB = L - B

Then,
R = L - CR
B = L - CB
G = CR + CB - L

The special color signals produced by the original demodulation scheme were cyan and magenta. Flesh tones are approximately equal to their sum. Broadcasting both cyan and magenta at equal energy levels minimizes the distortion of flesh tones.

RCA was able to recover CR and CB directly by the simple method of shifting the phase of the color carrier.

Have fun.

John R.
Post Reply