Siegel versus Arnott

Research on Safe Withdrawal Rates

Moderator: hocus2004

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

Siegel versus Arnott

Post by Mike »

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

Post by JWR1945 »

Thanks for an excellent article. This is my assessment.

I consistently found myself in Rob Arnott's camp. I disagree slightly on a couple of specifics.

1) On valuations, I much prefer reported earnings to estimated earnings. But I am reluctant to make any adjustments. We can see flaws in today's numbers, but how about yesterday's numbers? Were they really more accurate? I doubt it. [Of course, I prefer Dr. Shiller's P/E10 to a single year's P/E ratio.]
2) I consider the issue of accelerating earnings growth as unresolved. It will take several decades to get a definitive answer. I can point to an increased slope in the 1990s, but what does that mean? Others can make other comparisons that lead to the opposite conclusion. IMHO, neither is definitive. OTOH, the real earnings growth should be near 1.1% [Rob Arnott's number] or 1.2%[my number], but not anything nearly as high as 3.5%.

The critical remaining issue is supply and demand. I will defer to Mike's opinions at this time. I expect that there will be higher levels of selling in the future. But to some extent, I see many stocks being transferred to the younger generation as the older passes away. It is not as if everybody will make sure to have zero assets at the time of his death.

Have fun.

John R.
User avatar
Alec
Admin Board Member
Posts: 31
Joined: Wed Sep 10, 2003 4:00 am
Location: Crofton, MD

myths and such

Post by Alec »

Forecasting the equity premium is a competitive activity in the academic and practitioner world. The bulls have history on their side. Whether you look at the last 50, 100, or 200 years, real stock returns have been surprisingly constant, registering about a 7 percent annualized gain after inflation.

The amount by which they beat real bond returns have averaged between 4.5 percent and 5.5 percent over the last 75 years, according to Siegel. The optimists wonder why that should change.
How's this for a good reason for the equity premium to change: Everyone now knows that stocks "provide the highest long-term returns" - once everyone knows this, and buys stocks, their prices go up, and future returns go down. When everyone starts realizing that stocks are good long-term deals, they will cease to be such good long-term deals (like right now for example).

I also have to disagree slightly (or perhaps greatly) with the "real stock returns have been surprisingly constant, registering about a 7 percent annualized gain after inflation". As John's earlier posts (on this forum) showed, there was a wide variation in annualized returns over numerous time periods. Not to mention that fact that people actually only care about total returns, not annualized returns. If we look at the total returns, we find that there is not a convergence in returns, but rather a divergence.

I also agree with Arnott (and Bernstein - The Two Percent Dilution). I'd rather plan for lower returns, and be surprised by higher returns, than plan for higher returns, and be surpised by lower returns.

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

Post by JWR1945 »

Well done. I commend you, Alec, for addressing cause and effect, not numbers in isolation.

Here is additional reference material. I have bolded a comparison that drives Alec's point home.

These are the nominal returns (i.e., annualized return0) after 10, 20, 30, 40, 50 and 60 years when dividends are reinvested. Expenses are set to 0.00%.

Code: Select all

Start   10     20     30     40     50     60
1871   9.58   6.10   6.80   6.96   6.31   7.52
1921  13.78   7.91   9.62  11.00  10.34   9.93
1931   2.34   7.59  10.09   9.50   9.18   9.88
1971   7.90  10.63  12.80         
1981  13.44  15.33            
1991  17.26       


These are the real returns (i.e., annualized return0) after 10, 20, 30, 40, 50 and 60 years when dividends are reinvested. Expenses are set to 0.00%.

Code: Select all

Start   10     20     30     40     50     60
1871  12.69   8.61   8.53   7.76   5.42   7.09
1921  15.83   9.53   8.56   9.76   8.72   7.18
1931   3.57   5.10   7.80   7.02   5.53   6.03
1971  (0.22)  4.09   7.39         
1981   8.59  11.41            
1991  14.29               
These are some numbers that I have extracted from Professor Shiller's tables (relating to the S&P500 and/or its equivalent):

1871.01 (i.e., January 1871), P = 4.44, D = 0.26, E = 0.40, Real P = 60.130642, CPI = 12.464061

1921.01, P = 7.11, D = 0.5058, E = 0.7575, Real P = 63.166737, CPI = 19.0

1931.01, P = 15.98, D = 0.9667, E = 0.94, Real P = 169.64931, CPI = 15.9

1971.01, P = 93.49, D = 3.13, E = 5.16, Real P = 396.51035, CPI = 39.8

1981.01, P = 133, D = 6.20, E = 14.74, Real P = 258.05057, CPI = 87.0

1991.01, P = 325.49, D = 12.1067, E = 21.1833, Real P = 408.19251, CPI = 134.6

2001.01, P = 1335.63, D = 15.87, E = 48.48, Real P = 1287.5748, CPI = 175.1

I used the prices from 1871, 1921 and 1971 to check my calculator results after 50 years when no dividends were reinvested. Both real and nominal returns were correct.

These are the nominal returns (i.e., annualized return0) after 10, 20, 30, 40, 50 and 60 years when dividends are excluded. Expenses are set to 0.00%.

Code: Select all

Start   10     20      30     40     50     60
1871   3.38   0.43   1.56   1.86   0.95   2.16
1921   8.44   1.99   3.71   5.46   5.29   5.00
1931  (4.07)  1.43   4.49   4.52   4.33   5.15
1971   3.59   6.44   9.27         
1981   9.36  12.23            
1991  15.16               
These are the real returns (i.e., annualized return0) after 10, 20, 30, 40, 50 and 60 years when dividends are excluded. Expenses are set to 0.00%.

Code: Select all

Start   10     20      30     40     50     60
1871   6.31   2.81   3.20   2.63   0.10   1.74
1921  10.38   3.53   2.71   4.28   3.74   2.37
1931  (2.91) (0.92)  2.33   2.15   0.84   1.47
1971  (4.20)  0.15   4.03         
1981   4.69   8.40            
1991  12.24               
Compare the 50-year returns starting in 1921 and 1931. Shifting the starting date by a decade causes the real, long-term annualized return to change dramatically. This is true with and without dividend reinvestment.

IMHO, the 1921 and 1931 comparison is appropriate when looking for a prebubble and post bubble combination. The periods are different, yet similar. I do not expect another depression. OTOH, the bubble reached valuations far higher than those of the 1920s.

So much for the notion that a decade of bad prices can be ignored.

As a final item, I have calculated the real return (i.e., annualized return0) with dividends reinvested but with 0.20% expenses. As happens frequently, the percentages do not add or subtract exactly. In this case the differences are minor. [The tables that I have posted previously included 0.20% in fees.]

Code: Select all

Start   10     20      30     40     50     60
1871  12.47   8.40   8.31   7.55   5.21   6.87
1921  15.60   9.31   8.34   9.54   8.51   6.97
1931   3.37   4.89   7.59   6.80   5.32   5.82
1971  (0.42)  3.88   7.17         
1981   8.38  11.18            
1991  14.06   



Have fun.

John R.
Post Reply