Putting the CAPM to the test

1
Putting the CAPM to the test

• Black Jensen and Scholes (1972)
• Fama and McBeth (1973)
• Roll (1977)
• Fama and French (1992)

2
Whats the big idea?

• Deriving the CAPM involves two main steps
• The Market is Mean-variance efficient

• Because of this, everyone diversifies
• Therefore, idiosyncratic risk doesnt matter
• A stocks expected return depends only upon its
  beta

3
Two ways to test this

• First, the CAPM says that Beta matters in the
  following way.
• Check to see if, in the data, beta actually
  matters!
• Do higher beta stocks give you higher returns?
  How much more?
• Black, Jensen and Scholes (1972)

4
Black, Jensen and Scholes
5
Two ways to test this

• Second, the CAPM says that nothing else matters
• Fama and McBeth (1973) run a horserace between
  beta and total risk.
• i.e. use a stocks beta and its total risk to
  explain returns
• Consistent with CAPM, beta works better than

6
Fama and French

• Put size and book-to-market in the regression, as
  well as traditional market beta
• CAPM usually looks pretty bad.
• Depending on the period you study, and the
  methodology, the effect of beta sometimes
  actually goes the wrong way!

7
Rolls Critique

• The proxy we usually use in these tests is the
  stock market
• But the theory says it should be the diversified
  basket of all possible investable assets
• Corp. bonds, private equity, real estate, human
  capital.?

• If we dont have (cant get?) the right proxy,
  then the theory is neither testable nor useful

8
The future

• In the future the CAPM my be rescued or may be
  killed off.
• There are versions/extensions of the CAPM (ICAPM,
  CCAPM, etc.) that may prove more useful.
• The replacement for the CAPM will probably borrow
  some intuition / methodology from the plain
  vanilla CAPM.
• Or, better data / better proxies may make the
  CAPM more useful.
• Or, just use the Fama-French model to estimate
  expected returns (my suggestion at present)