Out of sample SWR (non-US)

Research on Safe Withdrawal Rates

Moderator: hocus2004

adrian2
* Rookie
Posts: 7
Joined: Sat Oct 23, 2004 9:40 pm

Post by adrian2 »

Mike wrote:I read the links. It would seem likely that SWRs for several countries are likely to be lower than US ones, based upon their lower equity returns.
I don't think lower equity returns it's the cause, it's the sequence of them. We only have one history for the US stock market, with not so many non-overlapping periods. Studying the way history unfolded in other places may give us a hint of what the future may have in hold for US, other than the usual "Stocks for the long run" optimism.

Adrian
Mike
*** Veteran
Posts: 278
Joined: Sun Jul 06, 2003 4:00 am

Post by Mike »

Bad things can certainly happen.
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

adrian2 wrote:I don't think lower equity returns it's the cause, it's the sequence of them. We only have one history for the US stock market, with not so many non-overlapping periods.
I have addressed this in my May 2004 Overview dated Wed May 19, 2004. It has a sticky and is near the top.
http://nofeeboards.com/boards/viewtopic.php?t=2505
Recently, we have found that earnings yield (using the average of the past decade's earnings) does an excellent job of predicting Safe Withdrawal Rates. It is even better than using dividend yield (plus about 1%) as a lower bound. The earnings yield overcomes the problem of surprise cuts in dividend amounts. It automatically corrects for any unsustainable payouts.
We already have a spin-off from our research, which I call A New Tool. It allows you to translate an assumed rate of return (perhaps for the next decade) into a Safe Withdrawal Rate (complete with confidence limits). It has an interesting side benefit. We can calculate how much a Safe Withdrawal Rate depends upon the exact sequence of returns as opposed to their overall total.

A New Tool: Overview from Wed Apr 28, 2004
http://nofeeboards.com/boards/viewtopic.php?t=2426
A New Tool from Wed Apr 28, 2004
http://nofeeboards.com/boards/viewtopic.php?t=2427

A New Tool goes into some depth about the effects of the sequence of returns and the overall stock market return. It provides numbers.

Here is another excerpt from my Overview post:
The worst case time period for starting a retirement was 1959-1973, not during the Great Depression. Stock dividends were very low by historical standards.
The worst case 30-year sequences in the United States started in 1965 and 1966, which is similar to the United Kingdom example.

Mike is right to point to market returns. They are related to earnings yield. Particular outcomes (Historical Surviving Withdrawal Rates) will vary, depending upon luck, which is the result of the sequence of returns. In the United States, we have been lucky. Safe Withdrawal Rates have been significantly lower than Historical Surviving Withdrawal Rates.

Another point: we distinguish between Historical Surviving Withdrawal Rates (also called Historical Database Rates) and Safe Withdrawal Rates. A Safe Withdrawal Rate is a lower confidence limit placed around calculations derived from and estimating Historical Surviving Withdrawal Rates.

Have fun.

John R.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

"The worst case 30-year sequences in the United States started in 1965 and 1966, which is similar to the United Kingdom example."

I'd like to run my understanding of this past you, JWR1945, and see whether we are working from the same conceptual framework or not.

My understanding is that there are two things that affect long-term returns: (1) the valuation level that applies at the beginning of a returns sequence; and (2) the pattern of ups and downs of the returns experienced during the returns sequence. These two factors combined produce the return (and the HSWR) for that historical sequence.

I accept that the sequence starting in 1965 turned out to be the least appealing sequence of returns. I do not believe, however, that the pattern of ups and downs of the returns experienced from 1965 onward was the worst pattern we have seen. It was a not particularly good (but also not particularly bad) pattern imposed on top of a particularly bad starting-point valuation level. Is that understanding of things in accord with your own?

"In the United States, we have been lucky. Safe Withdrawal Rates have been significantly lower than Historical Surviving Withdrawal Rates. "

I agree that we have never experienced the combination of a particularly bad valuation-level starting point plus a particularly bad pattern of returns that would produce a HSWR as low as some of the lowest SWRs we have experienced. However, I presume that we have experienced some poor returns patterns in historical sequences that started from not-so-bad valuation levels. That is, we have had "bad luck" both in terms of bad valuation starting points and in term of bad returns patterns but we have not yet had the double bad luck of having both forms of bad luck take place at the same time. Having both forms of bad luck take place at the same time would cause the proverbial worst-case scenario, one in which the HSWR would fall as low as the lowest of the SWRs on record in the United States.

Right now, we are at worst-case scenario conditions re the valuation starting-point factor. If we get a worst-case returns pattern on top of that, the HSWR for the current time will turn out to match the SWR for the current time. In the event that we get something better than a worst-case returns pattern, we will experience a HSWR higher than today's SWR.

The most likely possibility is that we will NOT get a worst-case scenario. My understanding of your research is that there is only a 5 percent possiblity that a 2.5 percent take-out will fail for a retirement starting today and financed by a portfolio of 80 percent S&P stocks. My further understanding is that there is a 50 percent chance that a 4.0 percent take-out wil fail for a retirement beginning at today's valuations. And that there is a 5 percent chance that even a take-out of 5.7 percent will work, presuming that stocks perform in the future somewhat in the way in which they have performed in the past.

Is the way that I am describing things here more or less on track?
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

Is the way that I am describing things here more or less on track?
You are doing fine.

For those interested in seeing some pictures, look at this graph from the special SWR Research section. It plots Historical Surviving Withdrawal Rates (also called Historical Database Rates) versus the percentage earnings yield 100E10/P (or 100% / [P/E10] ). The data are from sequences starting in 1923-1980.
http://www.nofeeboards.com/jwr/pic3.JPG

Technical background material is presented in the From Earnings Yield thread dated Thu Apr 15, 2004.
http://nofeeboards.com/boards/viewtopic.php?t=2368

I find it necessary to use the Full Screen mode in Windows. It is listed under View. You can also get the Full Screen mode by pressing key F11.

The lines are Calculated Rates. The lower confidence limit (relative to the Calculated Rate) is the Safe Withdrawal Rate.

Green and reddish brown colors are used with 50% stocks and 50% commercial paper. Data are presented as diamonds. Confidence limits are plus and minus 1.01%.

Blue and reddish purple colors are used with 80% stocks and 20% commercial paper. Data are presented as plus signs. Confidence limits are plus and minus 1.58%.

The Green line and the reddish brown diamonds describe what has happened with a 50% stock portfolio. Those diamonds below the green line are unlucky sequences. Safe Withdrawal Rates are 1% below the green line. One of those diamonds was unlucky enough to fall below the Safe Withdrawal Rate.

The reddish purple line describes what has happened with an 80% stock portfolio. Safe Withdrawal Rates are 1.6% below the reddish purple line. Those plus signs below the reddish purple line are unlucky sequences.

Notice that some sequences with low earnings yield are somewhat unlucky, but only slightly so.

Plots with Historical Surviving Withdrawal Rates (i.e., Historical Database Rates) show that the spread increases with earnings yield. This is awkward for statistics, but we do not claim great precision. We do claim consistency. Interestingly, the spread about similar plots using P/E10 instead of 100E10/P is well behaved, almost constant.

Have fun.

John R.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

JWR1945:

One of my favorite threads is the "Calculated Rates of the Past Decade" thread. In considering possible investment strategies, I find that it helps to have an idea of how quickly or slowly the SWR for stocks changes over time, and to what extent it changes.

Would you be able to provide the SWRs for 50 percent and 80 percent S&P portfolios for the years not covered in the "Calculated Rates" thread, going back to the early 1920s?
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

hocus2004 wrote:JWR1945:

One of my favorite threads is the "Calculated Rates of the Past Decade" thread. In considering possible investment strategies, I find that it helps to have an idea of how quickly or slowly the SWR for stocks changes over time, and to what extent it changes.

Would you be able to provide the SWRs for 50 percent and 80 percent S&P portfolios for the years not covered in the "Calculated Rates" thread, going back to the early 1920s?
This is what you have asked for. The equations are based on retirements starting in 1923-1980. I have applied the formulas using the percentage earnings yields from 1921-2003. I have used the full precision available in Excel in these updated calculations. This reduces some round-off errors.

Reference thread: Calculated Rates of the Last Decade dated Wed Jun 23, 2004.
http://nofeeboards.com/boards/viewtopic.php?t=2657
JWR1945 wrote:Calculated Rates of the Last Decade

I have applied the following equations from the thread From Earnings Yield dated Thu, Apr 15, 2004.
http://nofeeboards.com/boards/viewtopic.php?t=2368
With the 50% stock portfolio, the Historical Database Rate (HDBR50) equation is HDBR50 = 0.3979x+2.6434%, where x = 100*(E10/P) or 100/[P/E10] = the earnings yield in percent and R squared equals 0.6975. When using this equation, the standard deviation of HDBR50 is 0.6178. The 90% confidence limits are plus and minus 1.01% of the calculated value.

With the 80% stock portfolio, the Historical Database Rate (HDBR80) equation is HDBR80 = 0.6685x+1.6424%, where x = 100*(E10/P) or 100/[P/E10] = the earnings yield in percent and R squared equals 0.7274. The standard deviation of HDBR80 using this formula is 0.9649. The 90% confidence limits are plus and minus 1.58%.
I have listed below the last decade's January values of P/E10 taken from Professor Robert Shiller's website.
http://www.econ.yale.edu/~shiller/
http://www.econ.yale.edu/~shiller/data/ie_data.htm
Here are the values of P/E10 in January.

Code: Select all

1995    20.219819
1996    24.763281
1997    28.333753
1998    32.860928
1999    40.578255
2000    43.774387
2001    36.98056
2002    30.277409
2003    22.894158
Here are the Calculated Rates.
Year, P/E10, Percentage Earnings Yield 100E10/P, Calculated Rate for HDBR50, Calculated Rate for HDBR80

Code: Select all

1921    5.1    19.61   10.45   14.75
1922    6.3    15.87    8.96   12.25
1923    8.2    12.20    7.50    9.79
1924    8.1    12.35    7.56    9.90
1925    9.7    10.31    6.75    8.53
1926   11.3     8.85    6.16    7.56
1927   13.2     7.58    5.66    6.71
1928   18.8     5.32    4.76    5.20
1929   27.1     3.69    4.11    4.11
1930   22.3     4.48    4.43    4.64
1931   16.7     5.99    5.03    5.65
1932    9.3    10.75    6.92    8.83
1933    8.7    11.49    7.22    9.33
1934   13.0     7.69    5.70    6.78
1935   11.5     8.70    6.10    7.46
1936   17.1     5.85    4.97    5.55
1937   21.6     4.63    4.49    4.74
1938   13.5     7.41    5.59    6.59
1939   15.6     6.41    5.19    5.93
1940   16.4     6.10    5.07    5.72
1941   13.9     7.19    5.51    6.45
1942   10.1     9.90    6.58    8.26
1943   10.2     9.80    6.54    8.20
1944   11.1     9.01    6.23    7.66
1945   12.0     8.33    5.96    7.21
1946   15.6     6.41    5.19    5.93
1947   11.5     8.70    6.10    7.46
1948   10.4     9.62    6.47    8.07
1949   10.2     9.80    6.54    8.20
1950   10.7     9.35    6.36    7.89
1951   11.9     8.40    5.99    7.26
1952   12.5     8.00    5.83    6.99
1953   13.0     7.69    5.70    6.78
1954   12.0     8.33    5.96    7.21
1955   16.0     6.25    5.13    5.82
1956   18.3     5.46    4.82    5.30
1957   16.7     5.99    5.03    5.65
1958   13.8     7.25    5.53    6.49
1959   18.0     5.56    4.85    5.36
1960   18.3     5.46    4.82    5.30
1961   18.5     5.41    4.79    5.26
1962   21.2     4.72    4.52    4.80
1963   19.3     5.18    4.71    5.11
1964   21.6     4.63    4.49    4.74
1965   23.3     4.29    4.35    4.51
1966   24.1     4.15    4.29    4.42
1967   20.4     4.90    4.59    4.92
1968   21.5     4.65    4.49    4.75
1969   21.2     4.72    4.52    4.80
1970   17.1     5.85    4.97    5.55
1971   16.5     6.06    5.05    5.69
1972   17.3     5.78    4.94    5.51
1973   18.7     5.35    4.77    5.22
1974   13.5     7.41    5.59    6.59
1975    8.9    11.24    7.11    9.15
1976   11.2     8.93    6.20    7.61
1977   11.4     8.77    6.13    7.51
1978    9.2    10.87    6.97    8.91
1979    9.3    10.75    6.92    8.83
1980    8.9    11.24    7.11    9.15
1981    9.3    10.71    6.91    8.80
1982    7.4    13.48    8.01   10.66
1983    8.7    11.51    7.22    9.34
1984    9.8    10.25    6.72    8.49
1985    9.9    10.07    6.65    8.38
1986   11.7     8.57    6.05    7.37
1987   14.7     6.78    5.34    6.18
1988   13.8     7.28    5.54    6.51
1989   15.2     6.60    5.27    6.05
1990   17.0     5.88    4.98    5.57
1991   15.6     6.42    5.20    5.94
1992   19.6     5.11    4.68    5.06
1993   20.4     4.90    4.59    4.92
1994   21.5     4.65    4.49    4.75
1995   20.6     4.89    4.59    4.91
1996   25.4     3.93    4.21    4.27
1997   29.2     3.43    4.01    3.93
1998   33.8     2.96    3.82    3.62
1999   40.9     2.44    3.62    3.28
2000   44.7     2.24    3.53    3.14
2001   37.0     2.70    3.72    3.45
2002   30.3     3.30    3.96    3.85
2003   22.9     4.37    4.38    4.56
Here are the details for HDBR50.
Year, Safe Withdrawal Rate for HDBR50, Calculated Rate for HDBR50, High Risk Rate for HDBR50

Code: Select all

1921   9.44   10.45   11.46
1922   7.95    8.96    9.97
1923   6.49    7.50    8.51
1924   6.55    7.56    8.57
1925   5.74    6.75    7.76
1926   5.15    6.16    7.17
1927   4.65    5.66    6.67
1928   3.75    4.76    5.77
1929   3.10    4.11    5.12
1930   3.42    4.43    5.44
1931   4.02    5.03    6.04
1932   5.91    6.92    7.93
1933   6.21    7.22    8.23
1934   4.69    5.70    6.71
1935   5.09    6.10    7.11
1936   3.96    4.97    5.98
1937   3.48    4.49    5.50
1938   4.58    5.59    6.60
1939   4.18    5.19    6.20
1940   4.06    5.07    6.08
1941   4.50    5.51    6.52
1942   5.57    6.58    7.59
1943   5.53    6.54    7.55
1944   5.22    6.23    7.24
1945   4.95    5.96    6.97
1946   4.18    5.19    6.20
1947   5.09    6.10    7.11
1948   5.46    6.47    7.48
1949   5.53    6.54    7.55
1950   5.35    6.36    7.37
1951   4.98    5.99    7.00
1952   4.82    5.83    6.84
1953   4.69    5.70    6.71
1954   4.95    5.96    6.97
1955   4.12    5.13    6.14
1956   3.81    4.82    5.83
1957   4.02    5.03    6.04
1958   4.52    5.53    6.54
1959   3.84    4.85    5.86
1960   3.81    4.82    5.83
1961   3.78    4.79    5.80
1962   3.51    4.52    5.53
1963   3.70    4.71    5.72
1964   3.48    4.49    5.50
1965   3.34    4.35    5.36
1966   3.28    4.29    5.30
1967   3.58    4.59    5.60
1968   3.48    4.49    5.50
1969   3.51    4.52    5.53
1970   3.96    4.97    5.98
1971   4.04    5.05    6.06
1972   3.93    4.94    5.95
1973   3.76    4.77    5.78
1974   4.58    5.59    6.60
1975   6.10    7.11    8.12
1976   5.19    6.20    7.21
1977   5.12    6.13    7.14
1978   5.96    6.97    7.98
1979   5.91    6.92    7.93
1980   6.10    7.11    8.12
1981   5.90    6.91    7.92
1982   7.00    8.01    9.02
1983   6.21    7.22    8.23
1984   5.71    6.72    7.73
1985   5.64    6.65    7.66
1986   5.04    6.05    7.06
1987   4.33    5.34    6.35
1988   4.53    5.54    6.55
1989   4.26    5.27    6.28
1990   3.97    4.98    5.99
1991   4.19    5.20    6.21
1992   3.67    4.68    5.69
1993   3.58    4.59    5.60
1994   3.48    4.49    5.50
1995   3.58    4.59    5.60
1996   3.20    4.21    5.22
1997   3.00    4.01    5.02
1998   2.81    3.82    4.83
1999   2.61    3.62    4.63
2000   2.52    3.53    4.54
2001   2.71    3.72    4.73
2002   2.95    3.96    4.97
2003   3.37    4.38    5.39
Here are the details for HDBR80.
Year, Safe Withdrawal Rate for HDBR80, Calculated Rate for HDBR80, High Risk Rate for HDBR80

Code: Select all

1921  13.17   14.75    16.33
1922  10.67   12.25    13.83
1923   8.21    9.79    11.37
1924   8.32    9.90    11.48
1925   6.95    8.53    10.11
1926   5.98    7.56     9.14
1927   5.13    6.71     8.29
1928   3.62    5.20     6.78
1929   2.53    4.11     5.69
1930   3.06    4.64     6.22
1931   4.07    5.65     7.23
1932   7.25    8.83    10.41
1933   7.75    9.33    10.91
1934   5.20    6.78     8.36
1935   5.88    7.46     9.04
1936   3.97    5.55     7.13
1937   3.16    4.74     6.32
1938   5.01    6.59     8.17
1939   4.35    5.93     7.51
1940   4.14    5.72     7.30
1941   4.87    6.45     8.03
1942   6.68    8.26     9.84
1943   6.62    8.20     9.78
1944   6.08    7.66     9.24
1945   5.63    7.21     8.79
1946   4.35    5.93     7.51
1947   5.88    7.46     9.04
1948   6.49    8.07     9.65
1949   6.62    8.20     9.78
1950   6.31    7.89     9.47
1951   5.68    7.26     8.84
1952   5.41    6.99     8.57
1953   5.20    6.78     8.36
1954   5.63    7.21     8.79
1955   4.24    5.82     7.40
1956   3.72    5.30     6.88
1957   4.07    5.65     7.23
1958   4.91    6.49     8.07
1959   3.78    5.36     6.94
1960   3.72    5.30     6.88
1961   3.68    5.26     6.84
1962   3.22    4.80     6.38
1963   3.53    5.11     6.69
1964   3.16    4.74     6.32
1965   2.93    4.51     6.09
1966   2.84    4.42     6.00
1967   3.34    4.92     6.50
1968   3.17    4.75     6.33
1969   3.22    4.80     6.38
1970   3.97    5.55     7.13
1971   4.11    5.69     7.27
1972   3.93    5.51     7.09
1973   3.64    5.22     6.80
1974   5.01    6.59     8.17
1975   7.57    9.15    10.73
1976   6.03    7.61     9.19
1977   5.93    7.51     9.09
1978   7.33    8.91    10.49
1979   7.25    8.83    10.41
1980   7.57    9.15    10.73
1981   7.22    8.80    10.38
1982   9.08   10.66    12.24
1983   7.76    9.34    10.92
1984   6.91    8.49    10.07
1985   6.80    8.38     9.96
1986   5.79    7.37     8.95
1987   4.60    6.18     7.76
1988   4.93    6.51     8.09
1989   4.47    6.05     7.63
1990   3.99    5.57     7.15
1991   4.36    5.94     7.52
1992   3.48    5.06     6.64
1993   3.34    4.92     6.50
1994   3.17    4.75     6.33
1995   3.33    4.91     6.49
1996   2.69    4.27     5.85
1997   2.35    3.93     5.51
1998   2.04    3.62     5.20
1999   1.70    3.28     4.86
2000   1.56    3.14     4.72
2001   1.87    3.45     5.03
2002   2.27    3.85     5.43
2003   2.98    4.56     6.14
Have fun.

John R.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

That's wonderful. Thank you very much.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

"Studying the way history unfolded in other places may give us a hint of what the future may have in hold for US, other than the usual "Stocks for the long run" optimism. "

The words below are taken from William Bernstein's book "The Four Pillars of Investing. The appear in a section titled "Stock Returns Outside the U.S.," which begins on Page 29.

Bernstein: "U.S. stock returns of the past 200 years represent a best-case scenario. To get a more realistic view of stock returns, it's important to examime stock returns from as many nations, and over as long a period, as possible. Professors Philippe Jorion and William Groetzmann examined stock returns around the world in the 20th Century, and the picture they draw is not neary as pretty as the American story.

"Remember that, a century ago, the U.S. was an emerging market, and, two centuries ago, England, France, and Holland were also....It is a demonstration that the markets with the best returns survive, and that those with the worst returns do not--survivorship bias, yet again.

"The moral here is that because the most successful societies have the highest past stock returns, they become the biggest stock markets and are considered the most 'typical." Looking at the winners, we tend to get a distorted view of stock returns. It helps to recall that, three centuries ago, France had the world's largest economy, and just a century-and-a-half ago, that distinction belonged to England."
Post Reply