CHAPTER 9

T HE CA P M
BODI E , Z. , A . KA N E , A N D A . MA RCU S , 2 0 0 9. I N V ESTM ENTS .
8 T H E DI T ION. N Y: M CG R AW -HIL L/I RWIN.
THE CAPM AND THE INDEX MODEL
Is the CAPM testable even in principle?
To test the efficiency of CAPM (which predicts that the market portfolio is
a mean-variance efficient portfolio), considering that it treats all trade
risky assets, one would need to construct a value-weighted portfolio of a
huge size and test its efficiency.

2
THE CAPM AND THE INDEX MODEL
Is the CAPM testable even in principle?
Moreover, the CAPM implies relationships among expected returns,
whereas all we can observe are actual or realized holding-period returns
and these need not equal prior expectations.
To test mean-variance efficiency, we need to show that the reward-to-
volatility ratio is at its highest, but this ratio is set in terms of
expectations, which we cannot observe.
The expected return-beta relationship is also defined in terms of
expected returns 𝐸(𝑟 ) and 𝐸 𝑟 :
𝑅 =𝛼 +𝛽 𝐸 𝑟 −𝑟

3
THE INDEX MODEL AND REALIZED
RETURNS
The CAPM is a statement about ex ante or expected returns, but all
anyone can observe directly are ex post or realized returns.
To make the leap from expected to realized returns, one can employ
the index model (in excess return form):
𝑅 =𝛼 +𝛽𝑅 +𝑒
Turns out that this framework for statistically decomposing actual
stock returns meshes with the CAPM.

4
THE INDEX MODEL AND REALIZED
RETURNS
Turns out that this framework for statistically decomposing actual
stock returns meshes with the CAPM:
Start by deriving the covariance between the returns on stock 𝑖 and the
market index.
By definition, 𝐶𝑜𝑣(𝑅 , 𝑒 ) = 0.
It follows that
𝐶𝑜𝑣(𝑅 , 𝑅 )
= 𝐶𝑜𝑣 𝛽 𝑅 + 𝑒 , 𝑅 = 𝛽 𝐶𝑜𝑣 𝑅 , 𝑅 + 𝐶𝑜𝑣 𝑅 , 𝑒 = 𝛽 𝜎
( , )
Hence, 𝛽 =
The index model beta turns out to be the same beta as that of the CAPM
expected-return beta relationship, except that we replace the (theoretical)
market portfolio of the CAPM with the well-specified and observable
market index
5
THE INDEX MODEL AND THE EXPECTED
RETURN-BETA EQUATION
The CAPM expected return-beta relationship is, for any asset 𝑖 and
the (theoretical) market portfolio:
𝐸 𝑟 − 𝑟 = 𝛽 [𝐸 𝑟 − 𝑟 ]
This is a statement about the mean (or expected) excess of
returns relative to the mean excess return of the (theoretical)
market portfolio
Letting 𝑀 represent the true market portfolio in the index model and
taking expectations, the specification becomes:
𝐸 𝑟 − 𝑟 = 𝛼 +𝛽 [𝐸 𝑟 − 𝑟 ]

6
THE INDEX MODEL AND THE EXPECTED
RETURN-BETA EQUATION
Letting 𝑀 represent the true market portfolio in the index model and
taking expectations, the specification becomes:
𝐸 𝑟 − 𝑟 = 𝛼 +𝛽 [𝐸 𝑟 − 𝑟 ]
A comparison of the index model and the CAPM expected return-
beta relationship thus shows that the CAPM predicts that alpha
should be zero for all assets.
The alpha of a stock is its expected return in excess of (or below)
the fair expected return as predicted by the CAPM.
Hence, if a stock is fairly priced, its alpha must be zero.

7
THE INDEX MODEL AND THE EXPECTED
RETURN-BETA EQUATION
Note though that this is a statement about expected returns on a
security.
Ex post, some securities will do better or worse than expected with returns
higher or lower than predicted by the CAPM (i.e., they will exhibit positive or
negative alphas over a sample period).
The point is that this superior or inferior performance could not have been
forecast in advance.
If we estimate the index model for firms, the ex post or realized alphas
(intercepts) for the sample should center around zero.
The CAPM states that the expected value of alpha is zero for all securities.
The index model representation of the CAPM holds that the realized value of
alpha should average out to zero for a sample of historical observed returns.
Importantly, alphas should be unpredictable (i.e., independent from one
period to the next).

8
THE INDEX MODEL AND THE EXPECTED
RETURN-BETA EQUATION
There is another applicable variation on the index model, the market
model, which deals with "surprises":
𝑟 −𝐸 𝑟 =𝛽 𝑟 −𝐸 𝑟 +𝑒
If the CAPM is valid, this equation becomes identical to the index
model (try substituting 𝐸(𝑟 ))
Hence, the terms "index model" and "market model" are used
interchangeably

9
IS THE CAPM PRACTICAL?
Two key questions:
How can we determine if CAPM is the best available model to
explain rates of return on risky assets?
Even if it were the best theoretical model around to explain this,
how would it affect practical investment policy?

10
IS THE CAPM PRACTICAL?
From our discussion, we know that if the CAPM is valid, a single-index
model where the index contains all traded securities would also be valid.
Hence, if CAPM "perfectly" explained risky asset ROR, that would mean all
alpha values would be zero.
However, such a result cannot be expected to hold in real markets.
Grossman and Stiglitz (1981) showed that such an equilibrium may be one
the real economy can approach but never reach.
Basic idea: actions of security analysts drive prices to "proper levels" and
hence alphas to zero, — but if all alphas are identically zero, there would be
no incentive to engage in security analysis!
Instead, the market equilibrium would be characterized by prices hovering
"near" their proper values, at which alphas are almost zero, but with enough
slippage to induce analysts to continue their efforts (reward remains for
superior insight).
11
IS THE CAPM PRACTICAL?
A more reasonable standard for the CAPM as the "best available
model to explain the ROR of risky assets" is this:
In the absence of security analysis, one should take security
alphas as zero.
A security is mispriced if and only if its alpha is nonzero
(underpriced if positive and overpriced if negative), and positive or
negative alphas are revealed only by superior security analysis.
Absent the investment of significant resources in such analysis,
the an investor would obtain the best portfolio on the assumption
that all alpha values are zero.

12
IS THE CAPM PRACTICAL?
This definition of the superiority of the CAPM over any other model
also determines its role in real life investments.
Under the assumption that the CAPM is the best available model,
investors willing to expend resources to construct a superior
portfolio must:
First, identify a practical index to work with.
Deploy macro analysis to obtain good forecasts for the index .
Deploy security analysis to identify mispriced securities.

13
IS THE CAPM TESTABLE?
We can test a model in two ways:
Normative tests, which examine the assumption.
Positive tests, which examine the prediction.
In summary, preference for models that are "robust" with
respect to an assumption (i.e., predictions are not highly
sensitive to violation of the assumption), which will make the
model's predictions reasonably accurate.
Upshot is that tests of models are almost always positive (i.e.,
we judge a model on the success of its empirical predictions).

14
IS THE CAPM TESTABLE?
The CAPM has two main predictions (the second
derived from the first):
One, that the market portfolio is efficient.
Two, that the SML (the expected return-beta
relationship) accurately describes the risk-return tradeoff
(i.e., alpha values are zero).

15
IS THE CAPM TESTABLE?
Central problem in testing these predictions:
The hypothesized market portfolio, which includes all risky assets that can be
held by investors, is unobservable.
Apart from equities, such a portfolio should include bonds, real estate,
foreign asset, privately held businesses, and human capital (some of these are
traded thinly or not at all).
These problems make adequate testing of the CAPM infeasible.
Moreover, even small departures from efficiency in the market
portfolio could lead to large departures from the expected return-
beta relationship of the SML, negating the practical usefulness of the
model!

16
HOW WELL DOES THE CAPM DO IN
EMPIRICAL TESTS?
Because the market portfolio cannot be observed, tests of the CAPM
revolve around the expected return-beta relationship, using proxies
(market indexes) to stand for the true market portfolio.
The CAPM fails these tests.
The data reject the hypothesis that alpha values are uniformly
zero.
On average, low-beta securities have positive alphas, while high-
beta securities have negative alphas .

17
HOW WELL DOES THE CAPM DO IN
EMPIRICAL TESTS?
However, this is possibly a failure of: (i) data, (ii) the validity of the
market proxy, or (iii) the statistical method
Perhaps there is no better model out there, but we measure the
parameters (alpha and beta) with unsatisfactory precision (i.e., the
situation calls for improved technique).
But if rejection of the model is not due to statistical problems, useful
to search for extensions of the CAPM or substitute models.

18
HOW WELL DOES THE CAPM DO IN
EMPIRICAL TESTS?
CAPM is a widely accepted norm despite its empirical shortcomings:
◦ First, because the logic of the decomposition to systematic and
firm-specific risk (and especially the ability to assess nonmarket
components of risk premiums) is just so compelling
◦ Second, there is impressive, albeit less formal evidence that the
central conclusion of the CAPM—the efficiency of the market
portfolio—may be valid
The number of funds (those employing professional analysts and portfolio
managers and expending considerable resources to construct superior
portfolios) that consistently outperform a passive fund is small.
That is, a single-index model with ex ante zero alpha values may be a
reasonable working approximation for more investors.

19
HOW WELL DOES THE CAPM FARE IN
EMPIRICAL TESTS?
◦ The CAPM is widely used and accepted by investment industry
practitioners and the CFA, at least as a starting point for thinking
about the risk-return relationship.
When constructing optimal risky portfolios, practitioners must be satisfied
that the passive index they use for that purpose is satisfactory and that the
ratios of alpha to residual variance are appropriate measures of investment
attractiveness.
Many are reasonably satisfied with the estimation of alpha values and the
use of these values in constructing ORPs.

20
EXTENSIONS OF THE CAPM:
ZERO-BETA MODEL
The 'zero-beta' CAPM (capital asset pricing model) is a model that was
developed in response to the problems caused by the CAPM assumption
of unlimited borrowing and lending at the risk free rate.
Incorporated into the 'zero-beta' CAPM is a portfolio whose returns would
have absolutely no correlation with those of the market portfolio, i.e., the
'zero-beta' portfolio. This takes place of the risk-free asset.
A 'zero-beta' portfolio is constructed to have zero systematic risk or, in
other words, a beta of zero.
A zero-beta portfolio would have the same expected return as the risk-free rate.
Such a portfolio would have zero correlation with market movements, given that its
expected return equals the risk-free rate, a low rate of return.

21
EXTENSIONS OF THE CAPM:
ZERO-BETA MODEL
Efficient frontier portfolios have a number of interesting characteristics,
independently derived by Merton and Roll (1972), including the following:
Any portfolio that is a combination of two frontier portfolios is itself on the efficient
frontier.
The expected return of any asset can be expressed as an exact linear function of the
expected return on any two efficient-frontier portfolios P and Q according the
following equation:
𝐶𝑜𝑣 𝑟 , 𝑟 − 𝐶𝑜𝑣(𝑟 , 𝑟 )
𝐸 𝑟 − 𝐸 𝑟 = [𝐸 𝑟 − 𝐸 𝑟 ]
𝜎 − 𝐶𝑜𝑣(𝑟 , 𝑟 )
Every portfolio on the efficient frontier, except for the global minimum-variance
portfolio,  has  a  “companion”  portfolio  on  the  bottom  (inefficient)  half  of  the  frontier  
with which it is uncorrelated.
Because it is uncorrelated with the market portfolio, the companion portfolio is
referred to as the zero-beta portfolio.

22
EXTENSIONS OF THE CAPM:
ZERO-BETA MODEL
If we choose the market portfolio M and its zero-beta companion portfolio Z, then
the equation above simplifies to the CAPM-like equation:
𝐶𝑜𝑣 𝑟 , 𝑟 − 𝐶𝑜𝑣 𝑟 , 𝑟
𝐸 𝑟 −𝐸 𝑟 = 𝐸 𝑟 −𝐸 𝑟
𝜎 − 𝐶𝑜𝑣 𝑟 , 𝑟
𝐶𝑜𝑣 𝑟 , 𝑟
= 𝐸 𝑟 −𝐸 𝑟
𝜎
=𝛽 𝐸 𝑟 −𝐸 𝑟
This resembles the SML of the CAPM, except that the risk-free rate is replaced with
the expected return on the zero-beta companion of the market index portfolio.
Fischer Black showed that this is the CAPM that results when investors face
restrictions on borrowing and/or investment in the risk-free asset.
In this case, at least some investors will choose portfolios on the efficient frontier
that are not necessarily the market index portfolio.

23
EXTENSIONS OF THE CAPM:
ZERO-BETA MODEL
Because average returns on the zero-beta portfolio are greater than the
observed T-bill rates (the risk-free rate), the zero-beta model can explain why
average estimates of alpha values are positive for low-beta securities and
negative for high-beta securities, contrary to prediction of the CAPM.
Despite this, the model is not sufficient to rescue the CAPM from empirical
rejection!

24
EXTENSIONS OF THE CAPM:
LABOR INCOME AND NONTRADED ASSETS
An important departure from realism is the CAPM assumption that all
risky assets are traded
Two important asset classes that are not traded are: (i) human capital and
(ii) privately held businesses
The discounted market value of future labor income exceeds the total market
value of traded assets
The market value of privately held corporations and business is of the same
order of magnitude
Human capital and private enterprises are different types of assets with
possibly different implications for equilibrium returns on traded securities

25
 EXTENSIONS OF THE CAPM:
LABOR INCOME AND NONTRADED ASSETS
Of the two, privately held business may be the lesser source of
departure from the CAPM:
Nontraded firms can be incorporated or sold at will (save for liquidity
considerations)
Owners of private business can also borrow against their value, thus
diminishing the material difference between ownership of private and
public business
The CAPM expected return-beta equation may not be greatly disrupted by
the presence of entrepreneurial income if privately-held businesses have
similar risk characteristics as those of traded assets
Investors can partially offset the diversification problem posed by
entrepreneurial assets by reducing their demand for securities of the similar but
traded assets
26
EXTENSIONS OF THE CAPM:
LABOR INCOME AND NONTRADED ASSETS
To the extent that risk characteristics of private enterprises differ
from those of traded securities, however:
Excess demand for traded assets that best hedge the risk of
private business such that prices are bid up and expected returns
end up lower in relation to systematic risk.
Conversely, securities highly correlated with this risk will have
higher equilibrium risk premiums (and may appear to exhibit
positive alphas relative to the conventional SML).
Empirically shown how adding proprietary income to a standard
asset pricing model improves its predictive performance.

27
EXTENSIONS OF THE CAPM:
LABOR INCOME AND NONTRADED ASSETS
The size of labor income and its special nature is of greater concern for
the validity of the CAPM:
Despite the fact that an individual can borrow against labor income and
reduce some of the uncertainty about future labor income via life insurance,
human  capital  is  less  “portable”  across  time  and  may  be  more  difficult  to  
hedge using traded securities than nontraded business.
This may induce pressure on security prices and result in departures from the
CAPM expected return-beta equation.
Ex.  A  person  seeking  diversification  should  avoid  investing  in  his  or  her  employer’s  
stock and limit investments in the same industry; thus the demand for stocks of
labor-intensive firms may be reduced and these stocks may require a higher
expected return than predicted by the CAPM

28
EXTENSIONS OF THE CAPM:
LABOR INCOME AND NONTRADED ASSETS
Mayers (1972) showed that for an economy where individuals are endowed with labor
income of varying size relative to their nonlabor capital, the equilibrium expected
return-beta equation (and resulting SML) is
𝑃
𝐶𝑜𝑣 𝑅 , 𝑅 + 𝐶𝑜𝑣 𝑅 , 𝑅 Divide all terms in adjusted beta by
𝑃 sigma^2(m)
𝐸 𝑅 =𝐸 𝑅
𝑃
𝜎 + 𝐶𝑜𝑣 𝑅 , 𝑅
𝑃
where 𝑃 = value of aggregate human capital; 𝑃 = market value of traded assets
(market portfolio); 𝑅 = excess rate of return on aggregate human capital
The CAPM measure of systematic risk, beta, is replaced in this extended model by an
adjusted beta that also accounts for covariance with the portfolio of aggregate human capital.
Note that the ratio of human capital to market value of all traded assets, ,may well be
greater than 1, hence the effect of the covariance of a security with labor income,
𝐶𝑜𝑣 𝑅 , 𝑅 , relative to average, 𝐶𝑜𝑣 𝑅 , 𝑅 , is likely to be economically significant.

29
EXTENSIONS OF THE CAPM:
LABOR INCOME AND NONTRADED ASSETS
𝑃
𝐶𝑜𝑣 𝑅 , 𝑅 + 𝐶𝑜𝑣 𝑅 , 𝑅
𝑃
𝐸 𝑅 =𝐸 𝑅
𝑃
𝜎 + 𝐶𝑜𝑣 𝑅 , 𝑅
𝑃

When 𝐶𝑜𝑣 𝑅 , 𝑅 is positive, the adjusted beta is greater when the CAPM beta is smaller
than 1, and smaller when the CAPM beta is greater than 1
Hence, with 𝐶𝑜𝑣 𝑅 , 𝑅 likely to be positive for the average security, the risk premium in
this model will be greater, on average, than predicted by the CAPM for securities with beta
less than 1, and smaller for securities with beta greater than 1
This may help explain the average negative alpha of high-beta securities and the positive
alpha for of low-beta securities that lead to the statistical failure of the CAPM

30
 EXTENSIONS OF THE CAPM:
INTERTEMPORAL CAPM
Robert Merton revolutionized financial economics by using continuous-time
models to extend many of the asset pricing models.
In  his  basic  model,  he  relaxes  the  “single-period”  myopic  assumptions  about  
investors and envisions individuals who optimize a lifetime consumption and
investment plan and who continually adapt financial decisions to current
wealth and planned retirement age.
When uncertainty about portfolio returns is the only source of risk and
investment opportunities remain unchanged through time (i.e.,  there’s  no  
change in the probability distribution of the return on the market portfolio or
individual securities), his so-called intertemporal CAPM predicts the same
expected return-beta equation as the single-period equation.

31
 EXTENSIONS OF THE CAPM:
INTERTEMPORAL CAPM
However, the situation changes when we include additional sources of risk,
which are of 2 kinds:
First are changes in parameters describing investment opportunities (e.g., future
risk-free rates, expected returns, or the risk of the market portfolio).
Investors will sacrifice some expected return if they can find assets whose
returns will be higher when other parameters (e.g., the risk-free rate) change
adversely (in order to hedge risk).
Second are prices of the consumption goods that can be purchased with any
amount of wealth.
An example of an extramarket source of risk (apart from expected level and
volatility of nominal wealth) is inflation risk, and investors may be willing to
sacrifice some expected return to purchase securities whose returns will be
higher when the cost of living changes adversely (in order to hedge risk).

32
 EXTENSIONS OF THE CAPM:
INTERTEMPORAL CAPM
One can push this conclusion further, and argue that empirically significant
hedging demands may arise for important subsectors of consumer
expenditures (e.g., investors may bid up share prices of energy companies
that will hedge energy price uncertainty).
Effects that may characterize any asset that hedges important extramarket
sources of risk.

33
 EXTENSIONS OF THE CAPM:
INTERTEMPORAL CAPM
Suppose we can identify 𝐾 sources of extramarket risk and find 𝐾 associated hedge
portfolios,  then  Merton’s  ICAPM  expected  return-beta equation would generalize
SML to a multi-index version:

𝐸 𝑅 =𝛽 𝐸 𝑅 + 𝛽 𝐸 𝑅

where 𝛽 is the familiar security beta on the market index portfolio,

and 𝛽 is the beta on the 𝑘th hedge portfolio
Other multifactor models using additional factors that do not arise from
extramarket sources of risk have been developed and lead to SMLs of a form
identical to that of the ICAPM.

34
EXTENSIONS OF THE CAPM:
CONSUMPTION-BASED CAPM
The logic of the CAPM together with the hedging demands noted previously suggests
that it may be useful to center the model directly on consumption (models first
proposed by Mark Rubinstein, Robert Lucas, and Douglas Breeden).
In a lifetime consumption plan, the investor must in each period balance the allocation
of current wealth between today's consumption and the savings and investment that
will support future consumption.
When optimized, the utility value from an additional dollar of consumption today must
be equal to the utility value of the expected future consumption that can be financed
by that additional dollar of wealth
Future wealth will grow from labor income as well as returns on that dollar when
invested in the optimal complete portfolio
Wealth at each point in time being equal to the market value of assets in the balance
sheet plus the present value of future labor income
35
EXTENSIONS OF THE CAPM:
CONSUMPTION-BASED CAPM
Suppose risky assets are available and you wish to increase expected
consumption growth by allocating some of your savings to a risky portfolio—
how would we measure the risk of these assets?
As a general rule, investors will value additional income more highly during
difficult economic times (when consumption opportunities are scarce) than in
affluent times (when consumption is already abundant).
An asset will therefore be viewed as riskier in terms of consumption if it has
positive covariance with consumption growth (i.e., if its payoff is higher when
consumption is already high and lower when consumption is relatively
restricted).
Therefore equilibrium risk premiums will be greater for assets that exhibit
higher covariance with consumption growth.

36
EXTENSIONS OF THE CAPM:
CONSUMPTION-BASED CAPM
Developing this insight, we can write the risk premium on an asset a function of
its "consumption risk" as follows:
𝐸 𝑅 = 𝛽 𝑅𝑃
where:
Portfolio C may be interpreted as a consumption-tracking portfolio (also
called a consumption-mimicking portfolio), that is, the portfolio with the
highest correlation with consumption growth.
𝛽 is the slope coefficient in the regression of asset 𝑖's excess returns, 𝑅 ,
on those of the consumption-tracking portfolio.
𝑅𝑃 is the risk premium associated with consumption uncertainty, which is
measured by the excess return on the consumption-tracking portfolio.
37
EXTENSIONS OF THE CAPM:
CONSUMPTION-BASED CAPM
Consumption-based CAPM: 𝐸 𝑅 = 𝛽 𝑅𝑃
This conclusion noticeably very similar to the conventional CAPM, with both
approaches resulting in linear, single-factor models that differ mainly in the
identity of the factor they use.
The differences are:
 The consumption-tracking portfolio plays the role of the market portfolio in
the conventional CAPM.
This is in accord with its focus on the risk of consumption opportunities
rather than the risk and return of the dollar value of the portfolio.
The excess return on the consumption-tracking portfolio plays the role of the
excess return on the market portfolio, 𝑀.

38
EXTENSIONS OF THE CAPM:
CONSUMPTION-BASED CAPM
In contrast to the CAPM, the beta of the market portfolio on the
market factor of the CCAPM is not necessarily 1.
It is perfectly plausible and empirically evident that this beta is
substantially greater than 1.
This means that in the linear relationship between the market index
risk premium and that of the consumption portfolio:
𝐸 𝑅 =𝛼 +𝛽 𝐸 𝑅 +𝜀
where 𝛼 and 𝜀 allow for empirical deviation from the exact model
above, and 𝛽 not necessarily equal to 1.

39
EXTENSIONS OF THE CAPM:
CONSUMPTION-BASED CAPM
Just as the CAPM is empirically flawed, so is the CCAPM.
The attractiveness of this model is that it compactly incorporates
consumption hedging and possible changes in investment
opportunities (i.e., in the parameters of the return distributions) in a
single-factor framework.
There is a price to pay for this compactness, however: consumption
growth figures are published infrequently (monthly at most)
compared with financial assets, and are measured with significant
error.
Nevertheless, recent empirical research indicates that this model
is more successful in explaining realized returns than the CAPM.

40