Stock Market Forecasting

1
Stock Market Forecasting

• BKM 12.6 (pitifully short)
• Main reference the three articles linked on my
  webpage Campbell, Ritter, and Ibbotson

2
Two Benchmarks

• you ought to know.
• The Fed Model says the market fairly priced when
  the earnings yield (E1/P0) is equal to 10-year
  treasury yields.
• Lynchs model says that a security is fairly
  priced if its PEG ratio (P/E divided by 5-year
  year growth forecast) is less than one.
• These models are ad-hoc (i.e. not motivated by
  fundamental economics). They are theoretically
  wrong in that they compare real and nominal
  rates, do not control for risk, etc.)
• But they still may be useful

3

• Most common approach for individual securities
• Cannot work for forecasting the market itself
  (circular argument!). Instead use

Expected Premium on the Market
Current risk-free rate
4
Equity Premium Puzzle

• The historical data from 1926-present of the
  SP500 has a very high arithmetic rate of return.
• Though risk is higher than with bonds, there are
  very few long-run periods over which bonds win.
• Why would any long-term investor buy bonds?
• Possible Answer 1 Past returns were, in part, a
  surprise to investors. Future returns will not
  replicate past returns.

5
An Economic Argument

• Possible Answer 2 Historical risk premium should
  be higher than future risk premium
• Good old days had
• Larger transaction costs for entering the market
  fewer investors bearing the same amount of risk
• Less diversified basket of publicly traded
  securities
• Less available information (greater adverse
  selection problem?)
• (Usually) higher capital gains taxes

6
Future Risk Premium

• If some combination of Reason 1 and Reason 2 are
  valid, future risk premium should be less than
  historical risk premium of 7.
• Ibbotson forecast 10 in the 70s and he was
  right. Hes still forecasting 9 nominal, or
  about 7 real.
• So is the Social Security Advisory Board.
• Many others are shaving 3 off of this number.

7
John Campbells Calculation

• 1871-1997
• Div yield 5.4
• Geometric Stock Price Growth 1.6 (real)
• gt 7.0 real
• 1802-1997
• Div yield 4.9
• Geometric Stock Price Growth 2.1
• gt 7.0 real.

8
John Campells Calculation

• New Div Yield 1.4 (!!!)
• Dollars going to repurchases correspond to
  somewhat higher g than historical number (approx
  2)
• New real RETURN should be in the neighborhood of
  3.7 to 4.7.
• Since TIPS were yielding 3.5, equity premium is
  probably less than 1!

at the time Campbell was writing
9
Why is current D/P so low?

• Guess 1 Prices now are simply too high.
• D/P will correct over time as market returns are
  disappointing.
• Guess 2 Dividends now are temporarily low.
• High prices mean that the market forecasts high
  growth in earnings. Once this growth is
  realized, the ratio will return to normal.
• Ibbotson favors this explanation, and uses
  historical D/P in his 9 forever calculation.
• Guess 3 Structural Decline in Equity Premium
• You might argue that 1 actually makes sense from
  an economic perspective.
• (See again Possible Answer 2)

10
Campbells Middle-of-the-Road Position

• Campbell retreats from his back-of-envelope
  calculation that equity premium lt 1.
• Maybe all three explanations have some merit.
• Forecasts real geometric return of 5 to 5.5.
• Given TIPs return, this is a 1.5 to 2.0 equity
  premium.

11
Ritters Analysis

• Similar analysis as Campbell, but emphasizes
  interpreting g as growth rate in dividends rather
  than share prices.
• Argues that if dividends grow at a reasonable
  rate per capita, they will not eventually
  overwhelm labor income.
• Suggests g2.5
• D/P 1.5.
• gt Real return 4.0.
• Given TIPs return of 3.3, this is an equity
  premium of .7.

12
Who is Right?
Guess 2 D/P is temporarily low
Guess 1 Prices are too high
The Market Knows What its Doing
The Market Doesnt Know What its Doing