Gothenburg University

Financial Risk
MSA400

The Capital asset pricing

model and the Arbitrage
pricing theory

Authors:
Trang Nguyen
Olivia Stalin
Ababacar Diagne
Leonard Aukea
Supervisors:
Pr. Holger Rootzen
Dr. Alexander Herbertsson

This report has been written and analyzed by all the group
members jointly.

May 15, 2017

 The Capital Asset Pricing Model and the
Arbitrage Pricing Theory
Leonard Aukea, Ababacar Diagne, Trang Nguyen, Olivia Stalin

Abstract
In this work we review the basic ideas of the Capital Asset Pricing
Model and the Arbitrage Pricing Theory. Furthermore, we exhibit the
practical relevance and assumptions of these models. We show what
make them successful for the pricing of assets. Indeed, the drawback
and limitations of these models will be addressed as well.

Keywords: Capital Asset Pricing Model, Arbitrage Pricing The-

ory, asset pricing.

1 Introduction
Based on the pioneering work of Markowitz (1952) and Tobin (1958) for risky
assets in a portfolio, Sharpe (1964), Lintner (1965) and Mossin (1966) derived
a general equilibrium model for the pricing of assets under uncertainty, called
the Capital Asset Pricing Model (CAPM).
CAPM is a well-known and accepted single factor model, after four decades
CAPM is still one of the main alternatives in the estimation of expected
return or cost of equity for individual stocks (commodity derivatives, en-
ergy/electricity markets, etc.) and other financial securities. This task is
central to many financial decisions such as those relating to portfolio opti-
mization, capital budgeting, and performance evaluation. The measure of
risk in the CAPM is given by the security’s covariance with the market port-
folio, the so-called market beta.
Rather, the CAPM quantifies the expected rates of return of an asset with
its relative level of market systematic risk (beta). This explains why the

1
CAPM is called often a single factor model. Another model for the estima-
tion of asset returns is the Arbitrage Pricing Theory (APT). To improve the
discrepancy of the CAPM, the APT model was proposed by Stephen Ross
(1976) as a general theory of asset pricing. His theory predicts a relationships
between the returns of a single asset as a linear function of many indepen-
dent macroeconomic factors. In a historical context, the CAPM was the first
coherent framework answering the question of how expected returns and risk
were related.
In fact, as noted by authors in [15], the attraction of these model is that
it offers powerful and intuitively pleasing predictions about how to measure
risk and the relation between expected return and risk. The CAPM and APT
are simple asset pricing tools comparing to other probabilistic and stochastic
models.
Elsewhere, following authors in [12], the APT has generated an increased
interest in the application of linear factor models in the study of capital
asset pricing and a large academic literature [15, 13, 14, 10]. As an illustra-
tion some extensions and some modified versions of the CAPM have been
developed. For instance, the Conditional CAPM (CCAPM) models the time-
varying property of the distribution of stock returns (cf [18]). In addition, an
application of the consumption-based CAPM (pricing performance in seven
industry sub-sectors in the Taiwan stock market) is presented in [11]. Here,
the risk of a security is measured by the covariance of its return with per
capita consumption and is called consumption beta. Theoretically, the con-
sumption beta offers a better measure of systematic risk.
To name only a few, using stock listed in the Korean stock exchange,
authors in [21] present a comparative study by considering different versions
of CAPM and the APT models such as: CAPM, APT-motivated models, the
Consumption-based CAPM, Intertemporal CAPM-motivated models, and
the Jagannathan and Wang conditional CAPM model.
Besides, the CAPM and the APT have provided interesting and challeng-
ing research topics [17, 8, 5, 27, 26, 7]. For instance, Bartholdy and Peare
in [5] conducted a comparative study of the performance of this two models
for individual stocks. They make use of the Fama and French three-factor
model for the APT. As a result, they show that the Fama French model is
at best able to explain, on average, 5% of differences in returns on individ-
ual stocks despite the favor of the CAPM by practioners. Moreover, when
the risk levels are given by coherent risk measures such as VaR, CVaR, and
WCVaR, authors in [4] presented the optimal portfolio choice problems and

2
some extensions of the classic Arbitrage Pricing Theory (APT) and CAPM.
We refer the interested reader about these risk measures to [25, 16]. In [2],
the so called conditional CAPM is discussed. Likewise, by making use of
a Kalman filter, Tobias and Frazoni model the conditional beta and their
approach circumvents recent criticisms of this risk measure. They tackle a
number of the issues by assuming that betas change over time following a
mean-reverting process.
Moreover, Hwang and collaborators in [20] present a sharp idea of using the
credit spread as an option-risk factor and explain some limitations of the
traditional CAPM. They contribute to the option-risk CAPM literature by
using bond-credit-spread data as a proxy for default risk to control for the
option-risk characteristic of stocks. Their option-risk version of CAPM re-
sembles the conditional CAPM. Another application of the CAPM and APT
on the private equity asset class can be found in the recent paper [9]. Since
these asset by nature are illiquid, the traded liquidity factor is included in the
APT model. Even tough, an outstanding review and historical walk-through
of the CAPM is presented in [22]. The author claims that the CAPM devi-
ations is not due just to missing risk factors, hence the APT cannot be an
attempt to correct it.
Very recently, economically meaningful results with important policy im-
plications are found by By Aabo and co-authors in [1]. They introduce the
degree of the internationalization as a new corporate risk and illustrate their
results considered Scandinavian multinational firms. Hence, one can make
use of this factor in the APT.
Inter alia, Östermark in [3] revisited the portfolio efficiency of the APT and
the CAPM in Two Scandinavian Stock Exchanges (Finnish and Swedish).
As a finding, he demonstrates that the multifactor APT is more powerful in
predicting Finnish than Swedish stock returns, whereas the contrary holds
for the single factor CAPM.

Among others, the main aim of this present work is to present the basic
idea of the CAPM and the APT and we demonstrate the practical relevance
and assumption of these models show what make them successful for the
pricing of assets. Indeed, the drawback and limitations of these models will
be addressed as well.
The structure of this work is organized as follows. After the introduction
in first Section, we set up some preliminaries in Section 2. Section 3 is
devoted presentation of the CAPM and its derivation. The APT is discussed

3
in Section 4 before the conclusion and discussions in Section 5.

2 Preliminaries
To ease the understanding in the sequel we provide some concepts in finance
and definitions that will be of importance.

2.1 Markowitz portfolio theory

In the framework Markowitz mean-variance portfolio theory, the rate of re-
turns of a stock is considered as random variables. Assume, we would like to
build a compulsive portfolio from n given assets Ai ((ri ), σi2 ) where E(ri ) is
the expected return and σi2 the variance of the return. Here the variance of
the rate of return of an instrument is taken as a surrogate for its risk (volatil-
Xn
ity). On each asset Ai , we invest a proportion wi such that wi = 1. The
i=1
selection of an investment opportunity in the Markowitz theory is an opti-
mization problem. It consists of finding the optimal set of weights wi such
that the portfolio achieves an acceptable baseline expected rate of return un-
der a minimal volatility. Without loss of generality, in the Markowitz’ basic
principle investors will keep a risky security only if the expected return is
sufficiently high enough to compensate them for assuming the risk. This is
know as ‘Risk and Return trade-off’. We set the vector of investment pro-
portions w = (w1 , w2 , · · · , wn ) and we denote the covariance matrix between
the assets by C. The expected return of our portfolio is given by
X n
E(rP ) = wi E(ri ) (1)
i=1

and its variance

σ 2 (rP ) = wT Cw (2)
An optimal portfolio will be any investment strategies solving the problem
 1
 min wT Cw
2




subject to E(rP ) ≥ rb
 X n



 wi = 1.
i=1

4
2.2 Efficient Frontier
In a risk-return framework, a Markowitz Efficient Frontier, is the curve that
shows all the best combinations of securities that yield the maximum ex-
pected return for a given level of risk. It is defined also as the set of portfolios
that minimizes the risk subject to a given expected return. We are looking for
the investment policy that solves the previous optimization problem. Since
investors are generally risk averse, they will always invest in an efficient port-
folio. The efficient frontier (Pareto front?) traces out a increasing curve in
the risk-return plane as depicted here after in Figure 1.

Figure 1: The efficient frontier (EF) with two risky assets (left) or many
risky assets (right). The Capital market line (CML)tangent to the EF curve
gives the optimal portfolio (source [23]).

The portfolio on the EF that has the highest Sharpe Ratio, which will
be the point where the CML is just tangent to the EF will determine the
optimal portfolio.

2.3 Arbitrage
An arbitrage portfolio is a portfolio whose value V (t) satisfies the following
properties, for some time horizon T > 0 :
• V (0) = 0 almost surely,
• V (T ) ≥ 0 almost surely,
• P(V (T ) > 0) > 0 where P is a given probability measure.
In other words, a portfolio presents an arbitrage opportunity if it does not
require any initial wealth to hold it and guarantees a strictly positive return

5
at time T . Hence an arbitrage portfolio or simply arbitrage can be consid-
ered as a risk-free investment strategy. Basically an arbitrage represents a
deterministic money making instrument (which entails no risk).
We introduce this fundamental concept of arbitrage to ease the under-
standing in the sequel. In the playground of mathematics and finance, there
is a large number of work devoted to this concept of arbitrage. Nevertheless,
in sight of the complexity of modern markets arbitrage opportunities may
exist. They are characterized by practitioners as a result of market ineffi-
ciencies. The life time of an arbitrage is very short in time since they are
quickly exploited and traded away by investors, which actually stabilizes the
market. Without loss of generality, Björk in [6] interprets the existence of
arbitrage portfolio as a serious case of mispricing on the market. Let’s men-
tion that this mispricing is relative to a given asset pricing model.
With these pricing models included the CAPM, a common theoretical as-
sumption is the arbitrage-free principle which can be interpreted as: asset
prices in a financial market are such that no arbitrage opportunities can be
found. Its worthy to note that the arbitrage-free principle plays a key role
in finance and stand as the foundation of option pricing theory.

3 The Capital Asset Pricing model

As already stated in the introduction, the CAPM aims at a practical approach
to stock valuation. It is a model that gives you an appropriate expected rate
of return for each asset given the relevant risk characteristic of that asset
versus the market called beta. This latter factor beta measures the quantity
risk that cannot be diversified away. Like any other model, the CAPM is
constructed from a set of underlying assumptions about the real world. The
fundamental assumptions of the CAPM are as follows:

3.1 Assumptions of CAPM

In what follows, we review the assumptions behind the CAPM framework.
1. The financial markets are competitive
2. All investors plan to invest over the same time horizon
3. There is no distortionary taxes or transaction costs (markets are friction-
less)

6
 4. The investments are limited to publicly traded assets with unlimited
borrowing and lending at the risk-free rate and the market portfolio
consists of all publicly traded assets.

5. All investors like overall portfolio reward (expected return) and dislike
overall portfolio risk (variance or standard deviation of return)

6. Everyone either has quadratic utility or has homogeneous beliefs con-

cerning the distribution of security returns.

From these assumptions, one clearly sees that the CAPM is built under a
perfect competition assumption of microeconomics. The price of assets are
unaffected by the trades of investors which hold a small wealth compared
to the total endowment of all investors. We further observe that the total
return of any investor’s portfolio is a summation from two components: the
risk free assets and the risky market assets. This is due to the possibility of
lending and borrowing at the free rate. Furthermore, all the information is
available at the same time to all investors.
Besides its practical use, we note that the CAPM has many unrealistic as-
sumptions. For instance, the perfect competition assumption of microeco-
nomics does not hold, the forces of supply and demand determine the prices
of asset in reality between buyers and sellers. Since investments are limited
to a universe of publicly traded financial assets, this assumption rules out
many types of investments. Moreover, we know that investors are in differ-
ent tax brackets and that this may govern the type of assets in which they
invest. In other terms, naturally, there is no homogeneous expectations or
beliefs between investors. But this assumptions is crucial in the CAPM since
if the investors do not have similar expectations there will be no homogeneity
in their conception. Another highly unrealistic assumption is the fact that
investors should have identical time horizons, which obviously is not the case.
This assumption is a consequence of the CAPM being a single period model.
As an alternative, continuous time models are used to get over the above
difficulty of single periods. In summary, we may say that the these assump-
tions represent a very simplified and idealized world, nevertheless they are
crucial to arrive at the original and basic form of the CAPM as stated in the
following theorem.

Theorem 1 We denote by M the market portfolio with expected return E(rM )

2
and variance σM . We assume there exist a risk-free return rf .

7
For any asset i,
E(ri ) = rf + βi (E(rM ) − rf )
where
Cov(ri , rM )
βi = 2
σM
2
is called the beta of the asset i. The quantity σM is the variance of the market
portfolio.
More generally, for any portfolio p = (α1 , ..., αn ) of risky assets, its beta can
be computed as a weighted average of individual asset betas,

E(rp ) = rf + βp (E(rM ) − rf )

where n
X
βp = αi βi .
i=1

The beta value serves as an important measure of risk for individual assets
(portfolios) that is different from σi2 . It measures the non-diversifiable part
of risk called systematic risk. Besides, the CAPM can be seen as a single
factor model and the factor β can be appreciated as the factor sensitivity
of the asset’s return to the return of the market portfolio. To name only a
few, one aspect of the CAPM is that the expected return of an asset does
not depend on its stand-alone risk but depends on its sensitivity towards
the market. Another worth noting aspect of the CAPM is its role in testing
whether stock markets are efficient in terms of expected return.

3.2 Derivation of the CAPM

This subsection is devoted to the derivation of the CAPM. There are sev-
eral methods to arrive at the CAPM. Here, we follow ideas presented in [19]
that use an algebraic approach based on the Markowitz (financial engineering
mean-variance portfolio theory from Markowitz) portfolio theory. We con-
sider the market portfolio (M ) and a risky asset (Ai ). Investing a fraction
y of my investment funds in the the risky asset (Ai ) and the fraction 1 − y
in the market portfolio, the expected return and variance of my compulsive
portfolio (P = (M, Ai )) writes as:

E(rp ) = yE(ri ) + (1 − y)E(rM ) (3)

8
 σp2 = y 2 σi2 + (1 − y)2 σM
2
+ 2y(1 − y)σiM (4)
These quantities are a function of the investment proportion in the risky
asset y. Taking the derivative of E(rp ) and σp with respect to y, we obtain:

∂E(rp )
= E(ri ) − E(rm ), (5)
∂y
∂σp 2yσ 2 − 2(1 − y)σM2
+ 2(1 − 2y)σiM
= p i 2 . (6)
∂y 2 2 2
2 y σi + (1 − y) σM + 2y(1 − y)σiM
Since all investors use identical analysis of the same universe of assets, in
general equilibrium the market portfolio (M ) already includes the risky asset
(Ai ) (according to its market value weights). Then the proportion y can be
interpreting as measuring excess demand for the risky asset. Under general
equilibrium, no general equilibrium will exist i.e we set y to zero. Hence the
partial derivatives rewrites as:

∂E(rp ) 
= E(ri ) − E(rm ), (7)
∂y y=0
2

∂σp  σiM − σM
= . (8)
∂y y=0 σM
The market price of risk, the equilibrium risk-return trade-off will therefore
be the ratio
∂E(rp )
∂y E(ri ) − E(rM )
∂σp
= 2
σM . (9)
σiM − σM
∂y

Consider an investment scenario in the risk-free asset (F) and the market
portfolio (M). Due to the capital market line (CML), we have a relation for
the expected return of the portfolio:

E(rM ) − rf
E(rp ) = rf + σp . (10)
σM
Here the equilibrium, the market price of a risk (risk-return trade-off) is given
by:
E(rM ) − rf
. (11)
σM

9
Seen that in equilibrium, the marginal price-of-risk for all assets must be
equal (for none arbitrage reason), we get by equating equations (9) and (11)
the following relation

E(rM ) − rf E(ri ) − E(rM )

= 2
σM . (12)
σM σiM − σM

From this latter relation, one obtain:

σiM
E(ri ) = rf + 2
(E(rM ) − rf ). (13)
σM
σiM
We define the coefficient βi = 2
and rewrite (13) as
σM

E(ri ) = rf + (E(rM ) − rf )βi . (14)

This linear relationship between risk and return known as security market line
(SML) is the traditional CAPM model derived by Sharpe in 1964. Figure
2 shows a plot of the SML. The beta of the asset (Ai ), βi , measures the
quantity of risk exposure in asset (Ai ) while (E(rM ) − rf ) is the market price
σiM
of risk and 2 (E(rM ) − rf ) determines the market risk premium.
σM
One can view βi as a measure of non-diversifiable risk, the part of risk
correlated with the market that one can not reduce by diversifying. This
measure is also called the market risk or systematic risk. Often, the stocks
in big name companies which are deeply related to the general market are
expected to have high beta values.
As explained in [24], it makes sense that the CAPM only rewards in-
vestors for their portfolio’s responsiveness to the market. The risk that can
be diversified away is often called idiosyncratic or specific risk. The part that
cannot be escaped is often referred to as systematic risk or market risk. Beta
is a measure of this risk. Thus, even efficient portfolios will be exposed to
covariance risk.

10
Figure 2: The security market line. We see clearly for the market portfolio
we have βM = 1.

For the financial analyst, constructing the market portfolio would be a

difficult and unrealistic task in sight of the number of stocks in a real financial
market. To overcome this, index funds (index tracker or mutual fund) have
been created as an approximation of the market portfolio. This index is
made up by the most traded (dominant) assets in the market. The assets
capture the essence of the given market M , in other terms it mirrors the
market performance. The most commonly known index funds, the S&P 500
Index Fund (Standard Poor’s 500-stock index). In the Stockholm Stock
Exchange, the OMX Stockholm 30 (OM XS30) is the stock market index It
is a capitalization-weighted index that consists of the 30 most-traded stock
classes (see also Nikkei225, DAX30, CAC40, etc).

3.3 Pricing assets

Let the price of an asset at t = 0 be P = X0 . We set the payoff at Q = X1
at t = 1, expected payoff to E(X1 ) = Q̄ at t = 1. We denote the expected
return r̄ = E(ri ). By definition we have

E(Xi ) − X0 Q̄ − P
r̄ = =
X0 P
When using r̄ as the discount rate, P can be expressed as,

Q̄
P = .
r̄ + 1

11
From the CAPM formula we have

E(ri ) = rf + β(r̄M − rf )

Hence we obtain:
Q
P =
1 + rf + β(r̄M − rf )
Q−P Q
When using r = = − 1 then
P P
1 2
Cov(r, rM ) = Cov((Q/P )−1, rM ) = Cov((Q/P ), rM ) = Cov(Q, rM ) = σM βi .
P
And since β = Cov(r, rM ), we obtain,

1  Cov(Q, rM ) 
β= 2
.
P σM

And inserting this in the CAPM formula gives us,

Q̄
P = 1 2
.
(1 + rf + P
(Cov(Q, rM )/σM )(r̄M − rf )

P expresses the price of the present value of an expected pay off.

3.4 Limitations of the CAPM

The CAPM model offers a linear relationship between the systematic risk
and expected return for assets. The availability of the inputs and its original
simplicity make the CAPM an attractive and well-accepted tool for estimat-
ing the expected return of securities. The fact that the model considers the
systematic risk β is one of its great aspects. Namely, systematic risk is an
important factor since it often cannot be completely alleviated. Notwith-
standing, the CAPM makes some non trivial postulates that drive to many
drawbacks in the real world.
Inter alia, the CAPM assumes the existence of a perfect financial mar-
ket, where there are no restrictions on investments in terms of income taxes,
transaction costs etc. Obviously, this is far from the truth. This lack of a per-
fect market may induce an additional risk to the investors when they bump
into market regulations. The CAPM model assumes unlimited borrowing

12
and lending of the risk free ratio rf , and also that rf has same rate for all
investors. In reality individual investors are not allowed to borrow and lend
with the same rate as the government. The postula may lead to serious issues
in the valuation. Beyond the difficulties to estimate β another issue of the
CAPM is related to the return of the market. In fact, a problem arises when
the market return at a given time has a negative value. Furthermore, the
return of the market is not a proper representation of a future market return.
To correct these drawbacks, a large amount of research have been conducted
and several necessary improvements of the CAPM have been proposed in
the literature. These adjustments yield to multi-factor model that consider
not just beta but many source of risk related to the asset. Among other,
the so-called arbitrage pricing theory (APT) is have been a very attractive
alternative of the CAPM.
In constrast to the CAPM, the APT uses a finite number of factors and
the expected return of an asset will be related to its exposure to each of these
factors. In addition, the APT has more flexible assumption requirements
than the CAPM. In the following section present a review of the APT.

4 The Arbitrage Pricing Theory

The CAPM, equilibrium model of asset pricing, asserts that securities have
different expected returns just because they carry distinct betas. Per contra,
there exists an alternative model of asset pricing that was developed by
Stephen Ross and has more flexible assumption requirements. This model
is known as Arbitrage Pricing Theory (APT). One primary assumption of
the APT entails the use of arbitrage portfolio. This latter can be seen as an
investment tactic that evolves a short position on a security at a high price
and a simultaneously long position of the same security or its equivalent at a
low price. In effect, whenever they are discovered, investors have an incentive
to take advantage of arbitrage portfolio and this is an essential logic in the
APT.
As a result, the APT is a more generalized version of CAPM that allows the
modeler to extend the CAPM by adding additional macroeconomic factors to
the model. The APT can be qualified as an ”open-source” model. Namely,
the reason is that the APT does not specify either the number or identity of
the factors that drive the expected returns of an asset.
Focused on the interest rate, inflation and other aspects of the economy

13
such as GDP growth, oil prices etc, a lot of research has been conducted in
order to identify potential factors. There exists several versions of the APT
in the literature. For instance we cite the Fama and French Three Factor-
Model [13] and in 2011 Chen, Novy-Marx and Zhang published an alternative
three-factor model [10], their model fixes some anomalies of the Fama French.
More recently (2015), Fama and French proposed in [14] a five-factor model
directed at capturing the size, value, profitability, and investment patterns in
average stock returns performs better than the three-factor model of Fama
and French [13]. These models present good performances and are equipped
with sharp economic intuitions.

4.1 Assumptions of the APT

At the difference of the CAPM, the APT does not assume that investors hold
efficient portfolios. However it holds three underlying assumptions which are,

1. Asset returns are explained by systematic factors.

2. Investors can build a portfolio of assets where specific risk is eliminated

through diversification.

3. No arbitrage opportunity exists among well-diversified portfolios. If

any arbitrage opportunity exists, they will be exploited away by in-
vestors.

The CAPM is not equipped to determine the current price of stocks. Its
only efficiency is to return a stock’s expected return.

4.2 The APT equation(s)

Unlike in the CAPM, the APT assumes that the return of an asset is gen-
erated by a multiple factors model. Each factor can be viewed as a specific
beta coefficient towards a specific risk premium. Several researchers have
investigated stock returns and have requested the use of three to five factors
in the APT. For instance Stephen Ross and collaborators have identified the
following factors:
• rate of inflation

• growth rate in industrial production

14
 • spread between long-term and short-term interest rate

• spread between high-grade and low-grade bond

Other authors recommend to use the rate of interest, rate of change in oil
price and even rate of growth in defense spending. Furthermore, the market
index is often used as a factor in some versions.
Therefore in the case of n factors, the APT equation writes as

E(ri ) = λ0 + β1i λ1 + β2i λ2 + · · · + βni λn (15)

in the n-factors model, λj are the factors and βji the corresponding sensitiv-
ities. For the purpose of calibration λ0 is usually take as the risk free rate
because it has no sensitivity. Equation (15) in many framework, is written
under the form:

E(ri ) = rf + β1i (δ1 − rf ) + β2i (δ2 − rf ) + · · · + βni (δn − rf ). (16)

Here rf denotes the risk free rate and we have n risk premiums with their
sensitivities (specific ”betas” of the stock i). Some studies show that these
factors move with the market portfolio. As already stated above, we pre-
sented some empirical models for the APT. For instance, the Fama-French
three factors model writes as:
i i i
E(ri ) = rf + βm E(rM − rf ) + βSM B E(SM B) + βHM L E(HM L), (17)

where

• E(rM − rf ) is the excess expected return of the market

• E(SM B) and E(HM L) stand respectively as the expected return of

size factor and the BE/ME factor.

Lets mention that in their design, they work with two variables that are
represented by two portfolios named small minus big (SMB) and high minus
low (HML). They also consider a book-to-market equity (BE/ME) factor.
For more details, we refer the reader to [13]. The three different ”betas” can
be estimated by running time series regressions on historical data. Moreover,
Chen and co-authors considered two additional factors on the CAPM. The
new factors are based on investment and profitability [11]:

15
 • Factor INV: builds on the returns of a portfolio including companies
with low investments less the returns of a portfolio including companies
with high investment, low-minus-high INV.

• Factor ROE: builds on the returns of a portfolio including comp anies

with high return-on-equity less the returns of a portfolio including com-
panies with low return-on-equity, high-minus-low ROE.

Their model includes the market excess return (MKT) and can be recast in
the following form:
i i i
E(ri ) = rf + βM KT E(M KT ) + βIN V E(IN V ) + βROE E(ROE). (18)

For more details about the design of the model, we refer to [11] and references
therein. However their methodology is similar to the Fama and French [13]
and corrected anomalies of the latter.
Even better, Fama and French propose in a recent paper [14] a five-factor
model. Theoritically this new model is better than the three-factor model of
Fama and French [13]. In effect, the five-factor model is directed at capturing
the size, value, profitability, and investment patterns in average stock returns
[14].
In sight of these versions of the APT, there is no specific factor in APT
model, at least ex ante. On top of that, a first main difficulty that arises is to
identify the factors for a particular stock. Identifying and quantifying each
of these factors is challenging and is not a trivial endeavour. Another barrier
of the APT comes from its essential assumptions regarding the existence of
an arbitrage portfolio. For this reason,the model will not prevail if there is
no opportunity of arbitrage in the market.

5 Conclusions
In this work, the basic ideas of the Capital Asset Pricing Model and the
Arbitrage Pricing Theory are presented. Furthermore, we exhibit the prac-
tical relevance and assumptions of these models and show what make them
successful for the pricing of assets. APT has an advantage over CAPM due
to its flexibility, however it is more difficult in application since the factors to
be used are very difficult to identify. While The CAPM emphasize efficient
diversification and neglects the unsystematic risk, the APT uses the naive

16
diversification upon the law of large numbers and therefore neglects essential
risks, which is a part of the systematic risk.

In fact, the application of both methods shall be seen from a critical

standpoint. As a matter of fact, the models holds relatively unrealistic as-
sumptions of the real world. APT assumes a linear relationship between the
rate of return and the risk premiums, i.e the linear relationship is insufficient
it gives us a poor outcome.
The unrealistic assumptions exhibits homogeneous expectations about the
assets return is intuitively contradictory. Successful investors indeed has a
potential in comprehending features not accounted for. Furthermore, if all
investors do think and act alike, they would likely create a ’bubble’ which
will inflate the asset price and reduces the risk inherent in the asset. Besides
this, the betas used in the methods are extracted from historical data and
are unstable through time, i.e it holds no power in explaining a certain future
scenario of the market.

Finally, albeit the unrealistic assumptions of the real world, the methods
in general give us an accommodating valuation in some sense. Furthermore,
it is worth mentioning that no theory is perfect and it is worthwhile to learn
from theory object to the criticism.
Also, in sight of the vast amount of data generated by the financial industry
nowadays and the developments in Machine Learning, a natural question
that arises is to initiate a bottom up reformation for the validation of some
financial models. The use of techniques such as deep learning, may allow us
to use hidden patterns that may be valuable in unifying models such as the
CAPM and the APT to forecast risk.

17
References
[1] Aabo, T., Pantzalis, C., Sørensen, H., and Toustrup, M. T.
Corporate risk and external sourcing: A study of scandinavian multina-
tional firms. International Business Review 25, 6 (2016), 1297 – 1308.

[2] Adrian, T., and Franzoni, F. Learning about beta: Time-varying

factor loadings, expected returns, and the conditional capm. Journal of
Empirical Finance 16, 4 (2009), 537 – 556.

[3] Astermark, R. Portfolio efficiency of apt and capm in two scandina-

vian stock exchanges. Omega 18, 4 (1990), 433 – 444.

[4] Balbás, A., Balbás, B., and Balbás, R. Capm and apt-like models
with risk measures. Journal of Banking Finance 34, 6 (2010), 1166 –
1174.

[5] Bartholdy, J., and Peare, P. Estimation of expected return: Capm

vs. fama and french. International Review of Financial Analysis 14, 4
(2005), 407 – 427.

[6] Bjork, T. Arbitrage Theory in Continuous Time, 3 ed. Oxford Uni-

versity Press, 2009.

[7] Booth, G., Martikainen, T., and Tse, Y. Price and volatility
spillovers in scandinavian stock markets. Journal of Banking Finance
21, 6 (1997), 811 – 823.

[8] Booth, L. Optimal portfolio composition and the capm. Journal of

Economics and Business 35, 2 (1983), 205 – 211.

[9] Buchner, A. Risk-adjusting the returns of private equity using the

capm and multi-factor extensions. Finance Research Letters 16 (2016),
154 – 161.

[10] Chen, L., Novy-Marx, R., and Zhang, L. An alternative three-

factor model. Unpublished working paper, University of Rochester
(2011).

[11] Chen, M.-H. Risk and return: Capm and ccapm. The Quarterly
Review of Economics and Finance 43, 2 (2003), 369 – 393.

18
[12] Connor, G., and Korajczyk, R. A. Chapter 4 the arbitrage pricing
theory and multifactor models of asset returns. Handbooks in Operations
Research and Management Science 9 (1995), 87 – 144.

[13] Fama, E., and French, K. Common risk factors in the returns on
stocks and bonds. Journal of Financial Economics 33, 1 (1993), 3–56.

[14] Fama, E., and French, K. A five-factor asset pricing model. Journal
of Financial Economics 116, 1 (2015), 1 – 22.

[15] Fama, E. F., and French, K. R. The cross-section of expected stock

returns. The Journal of Finance 47, 2 (1992), 427–465.

[16] Föllmer, H., and Schied, A. Convex measures of risk and trading
constraints. Finance and Stochastics 6, 4 (2002), 429–447.

[17] Guermat, C. Yes, the capm is testable. Journal of Banking Finance

46 (2014), 31 – 42.

[18] Ho, C., and Hung, C.-H. Investor sentiment as conditioning infor-
mation in asset pricing. Journal of Banking Finance 33, 5 (2009), 892
– 903.

[19] Hull, J. Options, futures, and other derivatives, 6. ed., pearson inter-
nat. ed ed. Pearson Prentice Hall, Upper Saddle River, NJ [u.a.].

[20] Hwang, Y.-S., Min, H.-G., McDonald, J. A., Kim, H., and Kim,
B.-H. Using the credit spread as an option-risk factor: Size and value
effects in capm. Journal of Banking Finance 34, 12 (2010), 2995 – 3009.
International Financial Integration.

[21] Kim, S.-H., Kim, D., and Shin, H.-S. Evaluating asset pricing mod-
els in the korean stock market. Pacific-Basin Finance Journal 20, 2
(2012), 198 – 227.

[22] MacKinlay, A. Multifactor models do not explain deviations from

the capm. Journal of Financial Economics 38, 1 (1995), 3 – 28.

[23] Perold, A. F. The capital asset pricing model. Journal of Economic

Perspectives 18, 3 (September 2004), 3–24.

19
[24] Sanders, M. The Capital Asset Pricing Model Theory, Econometrics,
and Evidence. Master’s thesis, Department of Business Studies, Aarhus
School of Business, Aarhus University, 2011.

[25] Stoyanov, S. V., Rachev, S. T., and Fabozzi, F. J. Optimal

financial portfolios. Applied Mathematical Finance 14, 5 (2007), 401–
436.

[26] Waldenstrom, D. Why does sovereign risk differ for domestic and
external debt? evidence from scandinavia, 1938–1948. Journal of Inter-
national Money and Finance 29, 3 (2010), 387 – 402.

[27] Zabarankin, M., Pavlikov, K., and Uryasev, S. Capital asset

pricing model (capm) with drawdown measure. European Journal of
Operational Research 234, 2 (2014), 508 – 517. 60 years following Harry
Markowitz’s contribution to portfolio theory and operations research.

20