7.9 CAPM – what went wrong?
CAPM has been subjected to a large number of empirical tests. Most of these have
sought to estimate the beta for a security by regressing the monthly returns (capital
gain plus dividend expressed as a fraction of the security price at the beginning of the
month) against returns from the market portfolio(as we did for Ace plc in Example 7.3).
Since the market portfolio is strictly unobservable (it contains some of every possible
capital investment), a surrogate is used. Typically this is some representative capital
market index (in the UK the Financial Times Stock Exchange All Share Index has been
used for this purpose). Betas tend to be calculated on the basis of monthly returns over
fairly protracted periods such as five years (60 months). As with the market portfolio,
the risk-free rate is not strictly observable – all investments have some risk. Short-term
UK or US government securities have tended to be used as a surrogate for the risk-free
rate, since these are generally accepted as being close to risk-free investments.
The tests have then gone on to assess whether or not CAPM explains the returns for
securities during the period following the one that was used to estimate the betas of
those securities.
In the relatively early days of CAPM, it was broadly the case that the various tests
seemed to support the validity of the model. This, in turn, justified the use of CAPM
in estimating the required return for real investment projects. As a result, CAPM
increasingly became the standard way that financial managers estimated discount
rates for NPV assessments or hurdle rates if using IRR.
More recently, however, evidence has pretty consistently emerged that seriously
calls the validity of CAPM into question. Fama and French (2004) reviewed the evid-
ence on CAPM over the decades and conclude that ‘CAPM’s empirical problems
invalidate its use in applications’.
7.9 CAPM – what went wrong?
All that we have considered so far in this chapter should lead us to ask what is the
matter with CAPM and why early tests of its validity tended to support it. The model
seems to be based on impeccable logic, but it does rely on some assumptions that are
clearly invalid. Also still more assumptions need to be made so that the model might
be tested. Here might lie some of the problems. The assumptions that have been
identified as being problematic are:
l Investors are only concerned with a security’s expected value and standard deviation. It has
been argued that investors are concerned with more about the security than these
measures. If this is true, beta cannot be a complete measure of risk.
l A representative stock market index is a reasonable surrogate for the market portfolio. It is
common practice to use the returns information from a recognised stock market
index, such as the FTSE all-share index, in place of the ‘market portfolio’. The prob-
lem with this is that the true ‘market portfolio’ contains more than just the shares
listed on the London Stock Exchange. It theoretically contains some of every pos-
sible investment on earth.
l Returns from short-term government securities are a reasonable surrogate for the risk-free
rate. It is not strictly possible to view a true risk-free rate, because no investment is
strictly risk-free.
l CAPM is an ‘expectations’ model. Strictly the output from the model is what is ex-
pected to occur. Tests of the model, of necessity, test the model on the basis of what
actually has happened.
