evaluate the role of the CAPM in the modern asset pricing literature Introduction

Critically evaluate the role of the CAPM in the modern asset pricing literature
Introduction
Since the Capital Asset Pricing Model (CAPM) was represented by Sharpe and Lintner,
discussions about its adequacy do not stop, mainly due to its empirical contradictions and strong
assumptions. But despite this, the modern asset pricing literature considers this model as
foundational, and even critics admit its essential role in development of asset pricing theory. For
example, Fama and French (2004) in their work recognize the model as a cradle of asset pricing
theory and say that through the years it is extensively applied by practitioners for estimating the
portfolio behaviour.
In this paper, I want to consider the role of the CAPM in the perspective of the profound
research works which either cast doubt on this model or advocate it. But all of them have
contributed to the asset pricing theory. Merton (1972) noticed that despite the crucial impact on
the non-academic and academic financial community (the hundreds research works are based on
the CAPM), the model is still subject to strong criticism from the theoretical and empirical sides.
The first part of my work contains a brief analysis of the CAPM methodology with
special attention on the positive and negative aspects of its theoretical base and on the extension
models which relax some conditions of the CAPM.
In the second part I consider empirical tests of the CAPM and scrutinize how the
empirical tests and results trigger the developing of new theories, hypothesis and models (for
instance, Fama and French’s 3-factor model).
1. Methodology
According to Bailey (2005) the CAPM is inherently based on the mean-variance model
(M-V) with additional assumptions and in turn, mean-variance is established on the expectedutility theory (EUT). Levi (2010) claims that if “the foundations of EUT are not valid, further
implying that all models- including the M-V efficiency analysis and the CAPM, derived in the
EUT framework- are questionable” (p.44).
15/12/2010-15:39:41 <BE352-7-AU_10a1_1000583_65F8AB5835420378A78056D7B7E64702DD9CFE3F>
2
There are several well-known publications that put in doubt the validity of the expectedutility theory. For instance, Roy (1952) sharply criticizes the presumptions that the investors
always tend to maximise their return and always know the likely outcomes. In Levi’s (2010)
work, that I found very cognitive, the author refers to other theories that declare the expectedutility theory is indefensible. For example, he mentions the Prospect and Cumulative Prospect
theories that have had a big impact on financial economics development. However, the
researcher shows that even if expected utility theory in some cases is inadequate, “the CAPM is
surprisingly valid in behaviour economics and psychologists paradigms.” (p.68).
In compliance with Farell (1997) besides assumptions imposed by the mean-variance
theory, the CAPM should satisfy additional conditions. Those conditions are admitted to be
strong and unrealistic. That is why, in my view, scientists have two ways: either to relax
assumptions or to prove the validity of the CAPM with assumptions being violated.
For example, the presumption that the asset returns should be normally distributed is
statistically significant rejected. Levi (2010) names some papers dedicated to this issue, which
argue that despite of the normality rejection, the CAPM is valid, for instance, Levy and
Markowitz (1979) come to the conclusion that even if the normality is rejected the meanvariance model is valid and can be safely applicable; in addition Dunchin&Levy’s (2008) work
shows that financial loss despite of the normality rejection is negligible.
On the other hand, some presumptions can be relaxed by more complicated models.
Newman et al (1992) comment that in the early 70s scientists placed greater emphasis upon the
relaxing strong assumptions of the CAPM and achieved prosperous results.
The first extension model that I want to mention here is the zero-beta CAPM, which
removes the condition of unbounded borrowing and lending at the risk free rate. Brown &Reilly
(2009) note that there have been several studies dedicated to testing the zero-beta CAPM; some
of them support it, for instance, Stambaugh’s (1982) work; some rejected the model, for
example, Gibbons (1982) and Shanken (1985).
15/12/2010-15:39:41 <BE352-7-AU_10a1_1000583_65F8AB5835420378A78056D7B7E64702DD9CFE3F>
3
A significant development of the CAPM was made by Merton (1973), who adds the
intertemporal aspect in the classical CAPM (as we know the CAPM is single-period model).
This model finds its advocates as well as opponents, for example, Fama (1996) notes that the
“although Merton and Long show that the CAPM is a special case of the Intertemporal CAPM
(ICAPM), their formal treatments of the ICAPM lack the simple intuition that makes the CAPM
so attractive”(p.442). However, in 2004 Campbell&Vuolteenaho published the famous work
where they use the ICAPM as a basis of their own 2-beta model (I scrutinize this model in more
detail in the next part of the essay) and show that the model can sufficiently predict and explain
stocks returns.
Whereas the CAPM has strong and unrealistic assumptions and there have been several
attempts to relax them, I fully agree with Brown&Reilly (2009) that say “…theory should never
be judged on the basis of its assumptions but rather on how well it explains and helps us predict
behaviour in the real world. If this theory and the model it implies help us explain the rate of
return on a wide variety of risky assets, it is useful, even if some assumptions are unrealistic.”
(p.206).
2. Empirical findings and analysis
As Newman et al (1992) state the early efforts to explore the CAPM were mainly from
the theoretical side, but later researchers have become more concerned about empirical testing of
CAPM. The authors connected the beginning of this experimental stage with the Roll’s
publication (1977).
As Haugen (2001) notes, before Roll critiques there were several research works that
strongly supported the CAPM, for example Black et al (1972) test and Fama&MacBeth (1974)
research. In both works scientists use the similar methodology (they concentrate on the
characteristics of the security market line) and data (NYSE stock returns from the period 1926 to
1965).The only difference in two approaches is that F