SCV50 Tables

Research on Safe Withdrawal Rates

Moderator: hocus2004

Post Reply
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

SCV50 Tables

Post by JWR1945 »

SCV50 Tables

I have calculated the 30-year Historical Surviving Withdrawal Rates for a portfolio consisting of 50% Small Cap Value stocks and 50% commercial paper. Otherwise, these data were taken at conditions identical with those of HDBR50.

I used the Gummy 03 version of the calculator. It has Small Cap Value data for 1928-2000 that Gummy has collected in a database at his site.

Expenses were 0.20%. Portfolio's were rebalanced annually. Withdrawals were increased to match inflation in accordance with the CPI.

A portfolio would have had a positive balance at year 30 with withdrawals at the Historical Surviving Withdrawal Rate. The balance would have been zero or negative at a withdrawal rate that was bigger by 0.1%.

These tables include the Percentage Earnings Yield 100E10/P (which is 100 / [P/E10], where P/E10 is Professor Shiller's measure of valuation for the S&P500). They also include Historical Surviving Withdrawal Rates of HDBR50 for easy comparisons.

Have fun.

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

Post by JWR1945 »

Here are the tables.

Data for 1921-1927 use the standard S&P500 calculations from Professor Shiller's database. That is, SCV50 and HDBR50 are identical in those years. Years 2001-2010 have dummy data.

Only the sequences from 1928-1970 depend entirely on Small Cap Value data. Other sequences include some data from other sources.

Year. P/E10, 100E10/P, HDBR50, SCV50

Code: Select all

1921    5.1    19.61    8.1   8.2
1922    6.3    15.87    8.0   8.2
1923    8.2    12.20    7.5   7.8
1924    8.1    12.35    7.6   7.9
1925    9.7    10.31    7.3   7.7
1926   11.3     8.85    6.6   7.0
1927   13.2     7.58    6.4   6.9
1928   18.8     5.32    5.5   6.0
1929   27.1     3.69    4.5   5.1
1930   22.3     4.48    4.4   6.0
1931   16.7     5.99    4.5   7.4
1932    9.3    10.75    5.1   9.6
1933    8.7    11.49    5.7  10.8
1934   13.0     7.69    4.8   7.6
1935   11.5     8.70    5.2   7.9
1936   17.1     5.85    4.3   6.9
1937   21.6     4.63    3.9   5.3
1938   13.5     7.41    4.6   7.4
1939   15.6     6.41    4.4   6.9
1940   16.4     6.10    4.5   7.6
1941   13.9     7.19    5.4   9.5
1942   10.1     9.90    6.2  11.4
1943   10.2     9.80    6.1  11.1
1944   11.1     9.01    5.9   8.5
1945   12.0     8.33    5.7   7.4
1946   15.6     6.41    5.9   6.6
1947   11.5     8.70    7.1   7.9
1948   10.4     9.62    7.4   8.3
1949   10.2     9.80    7.3   9.0
1950   10.7     9.35    7.6   9.3
1951   11.9     8.40    7.1   8.3
1952   12.5     8.00    6.7   8.4
1953   13.0     7.69    6.5   8.6
1954   12.0     8.33    6.6   9.6
1955   16.0     6.25    5.6   7.8
1956   18.3     5.46    5.2   7.6
1957   16.7     5.99    5.3   8.0
1958   13.8     7.25    5.7   9.4
1959   18.0     5.56    4.9   7.6
1960   18.3     5.46    4.9   7.4
1961   18.5     5.41    4.8   8.1
1962   21.2     4.72    4.6   7.5
1963   19.3     5.18    4.8   8.4
1964   21.6     4.63    4.4   7.9
1965   23.3     4.29    4.2   7.6
1966   24.1     4.15    4.1   6.8
1967   20.4     4.90    4.4   7.6
1968   21.5     4.65    4.3   6.3
1969   21.2     4.72    4.3   5.5
1970   17.1     5.85    4.7   6.7
1971   16.5     6.06    4.8   6.9
1972   17.3     5.78    4.7   7.0
1973   18.7     5.35    4.7   7.6
1974   13.5     7.41    5.6   9.7
1975    8.9    11.24    6.7  11.9
1976   11.2     8.93    6.0  10.5
1977   11.4     8.77    6.2   9.3
1978    9.2    10.87    7.0   9.4
1979    9.3    10.75    7.3   9.9
1980    8.9    11.24    7.4   9.6
Have fun.

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

Post by JWR1945 »

A plot of SCV50 versus the percentage earnings yield 100E10/P was unsatisfactory when I used all of the results from 1921-1980. It had an R-squared value of 0.29. The problem had to do with 1921 and 1922 and their exceptionally high earnings yields.

Plots using 1928-1970 and 1928-1980 were satisfactory.

The equation of Calculated Rates using the 1928-1970 data is: y = 0.5392x + 4.1939 with an R-squared of 0.5969.

The equation of Calculated Rates using the 1928-1980 data is: y = 0.5591x + 4.113 with an R-squared of 0.6471. x is the percentage earnings yield. y is the Historical Surviving Withdrawal Rate estimated by a straight line curve fit.

Today's value of 100E10/P is around 3.5%. (P/E10 is around 28 to 29.) Using the 1928-1970 data, the Calculated Rate is 6.08%. Using the 1928-1980 data, the Calculated Rate is 6.07%.

In both cases, according to eyeball estimates, the upper confidence limit is plus 2.0% and the lower limit is minus 1.8%.

These are the rates:
Safe Withdrawal Rate 4.3%
Calculated Rate: 6.1%
High Risk Rate: 8.1%.

I caution readers that Small Cap Value might not actually exist. I found some links that bring up this issue:
The Truth about High Dividends dated Mon Aug 16, 2004.
http://nofeeboards.com/boards/viewtopic.php?t=2886
http://nofeeboards.com/boards/viewtopic ... 256#p23256
The next book on my reading list is What Works on Wall Street--Revised Edition, copyright 1998. by James P. O'Shaughnessy.
...
Another important point: the advantage claimed for small cap stocks applies only to micro-cap stocks (with capitalizations no higher than $25 million). Forget about that small cap value slice.
Dreman's Blockbuster dated Sun Sep 19, 2004.
http://nofeeboards.com/boards/viewtopic.php?t=2970

IIRC, David Dreman made an extensive review. He found that the Small Cap data that were being used by academia had serious flaws. These included extremely poor liquidity and prices reported as being midway between asked and offered regardless of whether any shares were traded.

Have fun.

John R.
Mike
*** Veteran
Posts: 278
Joined: Sun Jul 06, 2003 4:00 am

Post by Mike »

I caution readers that Small Cap Value might not actually exist.
That's interesting. I have had positive experiences with mutual funds that claim to be SCV since the fall of 2002, but they may not invest in the Fama/French data base stocks.

I am not sure a micro cap value index fund could grow large enough to justify its expenses, due to the small capitalization of the sector. Vanguard uses a statistical sampling approach to deal with small caps in their TSM fund, which tend to be larger than micros. Micro cap value seems to work best with individual client managers who can pick wisely.
Post Reply