PROGRAMMING FOR MACRO-FINANCE

Unit 3: II Capital Asset Pricing Model

Daniel Arrieta
[email protected]

IE University

Academic year: 24-25

Outline

Introduction

Portfolios with a risk-less asset

Efficient frontier calculation

Capital Asset Pricing Model

Risk and return in the CAPM

Annex: quadratic programming

Next

Introduction

Portfolios with a risk-less asset

Efficient frontier calculation

Capital Asset Pricing Model

Risk and return in the CAPM

Annex: quadratic programming

Introduction
In 1964 William Sharpe published Capital asset prices: A theory of
market equilibrium under conditions of risk.
Sharpe complemented Markowitz’s work and presented his model
Capital Asset Pricing Model popularly known as CAPM.
The most relevant contributions of Sharpe (1964) are:
i. The generalization of the efficient frontier by including the
risk-free asset. This new efficient frontier is called Capital Market
Line (CML).
ii. The definition of diversifiable and non-diversifiable risks. The
latter is also called systematic risk.
One year later, Lintner (1965a), derived the CAPM from the
perspective of a company that issues shares.
Mossin (1966) independently obtained the CAPM by using quadratic
utility functions.
1/31
Next

Introduction

Portfolios with a risk-less asset

Efficient frontier calculation

Capital Asset Pricing Model

Risk and return in the CAPM

Annex: quadratic programming

Next

Introduction

Portfolios with a risk-less asset

Risk free asset
CML and Sharpe ratio

Efficient frontier calculation

Capital Asset Pricing Model

Risk and return in the CAPM

Annex: quadratic programming

Risk free asset
Markowitz’s model assumed that σi2 > 0 for all i. In other words, all
the assets comprising the financial market carried risk.
It is possible to introduce an asset without risk, such as a zero coupon
bond issued by a state in its own currency and held in the portfolio
until maturity.
We will denote the random variable return of the risk-free asset by Rf ,
and for the sake of notational simplicity, its sub-index will be f .
The risk-free asset main properties are:
i. E [Rf ] =: r ,
ii. σf2 = 0, and
iii. ρi,f = 0 for each risky asset i.
The next step is taking into account that the market includes Rf , and
therefore it should be included in the composition of the analyzed
portfolios.
2/31
Portfolios with a risk free asset
In general, the portfolio that includes Rf can be regarded as a
portfolio with two assets, one risky asset and the Rf itself.
The risky asset will be a portfolio that contains a proportion of each
single risky asset k such that σk2 > 0.
The random variable return of the aforementioned generic risky asset
will be denoted by Ri .
Therefore the portfolio return is given by RP = wi Ri + (1 − wi )Rf ,
where wi denotes the weight of Ri in the portfolio.
RP expected return is given by

µP := E [RP ] = wi µi + (1 − wi ) r
(1)
= r + wi (µi − r ) ,
and its variance
h i
σP2 := E (RP − µP )2 = w 2 σi2 . (2)
3/31
Next

Introduction

Portfolios with a risk-less asset

Risk free asset
CML and Sharpe ratio

Efficient frontier calculation

Capital Asset Pricing Model

Risk and return in the CAPM

Annex: quadratic programming

Capital Market Line
Using the value of wi given implicitly in (2) in the expected return of
the portfolio given by (1) yields

µi − r
 
µP = r + σP . (3)
σi
This last relationship implies that when Rf is included, the portfolio
return’s µP is linear in its risk σP .
Definition (risky asset Capital Market Line)
Given an asset with risk Ri and another without risk Rf , the
relationship between the expected return of a portfolio RP composed
of both assets and its risk is given by the equation (3).
This linear relationship is said to be the Capital Market Line (CML)
of the considered risky asset Ri .
Usually it is called simply the Capital Market Line of the asset.

4/31
Sharpe Ratio
The slope of the Capital Market Line given by former equation (3) has
its own name.
Definition (Sharpe Ratio)
Given the Capital Market Line of the risky asset Ri , the slope of said
line is said to be the Sharpe Ratio or risk premium of Ri .
µi − r
 
SR (Ri ) := .
σi

Definition (Market portfolio)

The portfolio with the highest Sharpe ratio is said to be the market
portfolio or the optimal risky portfolio.
These last concepts will make sense by recalling the Markowitz’s
efficient frontier definition, and its corresponding plot in the (σ, µ)
plane.
5/31
Next

Introduction

Portfolios with a risk-less asset

Efficient frontier calculation

Capital Asset Pricing Model

Risk and return in the CAPM

Annex: quadratic programming

Next
Introduction

Portfolios with a risk-less asset

Efficient frontier calculation

including the risk free asset
Tobin separation theorem
Sharpe ratio maximization

Capital Asset Pricing Model

Risk and return in the CAPM

Annex: quadratic programming

Riskless asset and the efficient frontier
The following figure displays the efficient frontiers without including
the risk-free asset in blue and including it in green.

Figure 1: Efficient frontier including the risk-free asset (green line) and not
including it (blue hyperbola).
6/31
Next
Introduction

Portfolios with a risk-less asset

Efficient frontier calculation

including the risk free asset
Tobin separation theorem
Sharpe ratio maximization

Capital Asset Pricing Model

Risk and return in the CAPM

Annex: quadratic programming

Calculation of the efficient frontier with a risk-free asset
To obtain the efficient frontier, the only portfolio composed solely of
risky assets that must be calculated is the market portfolio as seen on
the Figure 1.
The following theorem states how to proceed in such calculations.
Theorem 1 (Tobin separation theorem)
Given an investment universe that includes risk-free assets, and
assuming risk aversion with a quadratic utility function in decision
making, then the efficient portfolio selection process can be separated
into two independent steps:
i. Calculation of the market portfolio or optimal risky portfolio by
maximizing the Sharpe ratio.
ii. Obtaining the optimal efficient portfolio as a combination of the
market portfolio with the risk-free asset.

The proof of this theorem can be found in Tobin (1958).

7/31
Next
Introduction

Portfolios with a risk-less asset

Efficient frontier calculation

including the risk free asset
Tobin separation theorem
Sharpe ratio maximization

Capital Asset Pricing Model

Risk and return in the CAPM

Annex: quadratic programming

Maximizing the Sharpe ratio
The goal of maximizing the Sharpe ratio is given by the program

wT µ − r
maxn √ ,
w∈R wT Σw
(4)
s.t.
wT 1 = 1,
where w and µ are column vectors of dimension n that respectively
contain the weights of the risky assets and their expected returns.
The return of the risk-free asset is r , Σ denotes the covariance matrix
between the returns of the risky assets, and 1 is a column vector of
dimension n whose components are all equal to 1.
The optimization problem (4) is difficult due to the nature of the
objective function.
However, under certain assumptions, it can be reduced to a standard
quadratic program.
8/31
Equivalent quadratic program
The aforementioned optimization problem can be reduced to a
standard quadratic program as stated in the below
Proposition 2
If exists w ∈ Rn such that wT µ − r > 0 and wT 1 = 1, then
optimization problem (4) is equivalent to
1 T
minn z Σz,
z∈R 2 (5)
s.t. µ̂T z = 1,
where µ̂T := µ − r 1.
The hypotheses of the proposition assume that the universe of risky
assets is capable of generating a portfolio whose expected return is
higher than the expected return of the risk-free asset.
The proof of the equivalence between the programs (4) and (5) can
be found in the Annex.
9/31
Equivalent quadratic program (ii)
The resolution of the optimization program (5) is carried out using
the Lagrange multipliers technique.
The necessary condition for a local maximum, i.e., null gradient,
results in ! ! !
Σ µ̂ z 0
= , (6)
µ̂T 0 λ 1
where 0 is a column vector of dimension n whose components are all
equal to zero.
As detailed in the Annex, the z∗ that is a solution of (6) are not a
linear affine combination, i.e., the weights do not add up one, so the
solution of (4) is given by

1
w∗ = Pn ∗ z∗ , (7)
i=1 zi

where zi∗ denotes the i-th z∗ component.

10/31
Sharpe ratio maximization example
Considering the following estimation for the covariance matrix and the
expected return vector of the risky assets
! !
0.0275 −0.0075 0.05
Σ= , µ= .
−0.0075 0.0056 0.02

It is also known that the return of the risk-free asset is r = 0.01.

We want to compute the market portfolio composition.
First we calculate the system of equations given by (6), which in this
case is
    
0.0275 −0.0075 0.04 z1∗ 0
 ∗  
 −0.0075 0.0056 0.01   z2  =  0  .
 
0.04 0.01 0 λ∗ 1

11/31
Sharpe ratio maximization example (ii)
The last system of equations solution is given by
   
z1∗ 16.8831
 ∗  
 z2  =  32.4675  ,

λ∗ −5.5195

applying the normalization stated in equation (7) the weights of the

market portfolio are
! !
w1∗ 1 z1∗
wm := = ∗
w2∗ z1 + z2∗ z2∗
!
0.3421
= .
0.6579

An easy calculation yields that the volatility and the expected return
of the market portfolio are given by σm = 3, 03%, and µm = 4, 76%
respectively.
12/31
Next

Introduction

Portfolios with a risk-less asset

Efficient frontier calculation

Capital Asset Pricing Model

Risk and return in the CAPM

Annex: quadratic programming

Next
Introduction

Portfolios with a risk-less asset

Efficient frontier calculation

Capital Asset Pricing Model

modeling assumptions
efficient portfolio return
risk-return ratio equilibrium

Risk and return in the CAPM

Annex: quadratic programming

Modeling hypothesis
The Capital Asset Pricing Model (CAPM) assumes:
i. The joint probability distribution of all risky assets returns is
known by all market participants.
ii. There are no transaction costs or taxes
iii. Assets are indefinitely divisible.
iv. You can invest in all assets without restrictions.
v. Investors maximize their expected utility using the mean-variance
criterion.
vi. There is a risk-free asset.
vii. Investors only have efficient portfolios.
viii. The supply and demand of financial assets is in equilibrium.
ix. The prices of financial assets are given and do not depend on the
investors.
x. The model is static and only considers one period of time
13/31
Implications of the hypotheses
The main implications of the CAPM model assumptions are
i. Each investor selects his efficient portfolio through the two
sequential and independent steps described in the Theorem 1
(Tobin separation theorem).
ii. All investors and market participants have the same market
portfolio.
The first assumption, see Sharpe (1964) and Lintner (1965a), is very
restrictive. Lintner (1965b) calls it the hypothesis of idealized
uncertainty.
Other assumptions about this first hypothesis have been analyzed by
Sharpe (1970), where he also considers different funding and
investment rates.
There are other assumptions of the model that are easily dropped,
such as limiting portfolios to convex linear combinations. The latter
was addressed by Black (1972).
14/31
Next
Introduction

Portfolios with a risk-less asset

Efficient frontier calculation

Capital Asset Pricing Model

modeling assumptions
efficient portfolio return
risk-return ratio equilibrium

Risk and return in the CAPM

Annex: quadratic programming

Efficient portfolio return
The random variable return of an efficient portfolio RP is given by
n
X
RP = Rf + ωi (Ri − Rf ) ,
i=1

where
ωi := ωm wi∗ .
In this setting ωm denotes the weight of the market portfolio in the
given efficient portfolio, i.e. it depends on the utility function of the
investor.
On the other hand, wi∗ stands for the weight of the i-th risky asset in
the market portfolio, and it is the same for all investors by hypothesis.
Hence the expected return of RP is
n
X
µP = r + ωi (µi − r ) . (8)
i=1
15/31
Marginal contribution to portfolio’s risk and return
The variance of RP is obtained from
n X
X n
σP2 = ωj ωk σjk . (9)
j=1 k=1

Definition (marginal contribution)

The partial derivative of µP respect to ωi , that is ∂µ P
∂ωi , is said to be
the i-th risky asset marginal contribution to the expected return
of the portfolio RP .
∂σ 2
Similarly, the partial derivatives ∂ωPi and ∂σ
∂ωi are called respectively
P

the i-th risky asset marginal contribution to the variance, and

marginal contribution to the standard deviation, of the portfolio
RP .
Recall that, in a little abuse of language, the expected return of a
portfolio is called just the return, and the the standard deviation is
just called the risk.
16/31
Marginal contribution to portfolio’s risk and return (ii)
Deriving both sides of equation (8) yields the marginal contribution to
the portfolio’s return
∂µP
= µi − r .
∂ωi
On the other hand, from (9) can be obtained the marginal
contribution to the variance as follows
n
∂σP2 X
=2 ωk σik
∂ωi k=1
= 2 Cov (RP , Ri ) .

Lastly, by means of the chain rule, the marginal contribution to the

portfolio’s risk is
∂σP 1 ∂σP2
=
∂ωi 2σP ∂ωi
σiP
= .
σP
17/31
Next
Introduction

Portfolios with a risk-less asset

Efficient frontier calculation

Capital Asset Pricing Model

modeling assumptions
efficient portfolio return
risk-return ratio equilibrium

Risk and return in the CAPM

Annex: quadratic programming

Equilibrium of the risk-return ratio
Definition (risk-return ratio)
Given the portfolio RP , it is said that the quotient

∂µP /∂ωi µi − r
= ,
∂σP /∂ωi σiP /σP

is the portfolio’s i-th risky asset risk-return ratio.

Proposition 3 (risk-return ratio equilibrium)

If the CAPM hypotheses listed above are fulfilled, then for each risky
asset i in the financial market it is verified
µm − r µi − r
= ,
σm σim /σm
where m denotes the market portfolio sub-index.
The proof of the proposition is detailed in Sharpe (1964).

18/31
Capital Asset Pricing Model
The main result from Proposition 3 (risk-return ratio equilibrium), is
given by rewriting it in the following manner
σim
µi − r = 2
(µm − r ) .
σm

This last equality gives rise to the following

Definition (beta)
Given the market portfolio Rm and the i-th risky asset, it is said that
the quotient σσim
2 is the the i-th risky asset beta regarding to the
m
market portfolio, and it is denoted by βi .
Using the βi , above equation can be written as

µi − r = βi (µm − r ) , (10)

which is the standard expression of the CAPM.

19/31
Next

Introduction

Portfolios with a risk-less asset

Efficient frontier calculation

Capital Asset Pricing Model

Risk and return in the CAPM

Annex: quadratic programming

Next
Introduction

Portfolios with a risk-less asset

Efficient frontier calculation

Capital Asset Pricing Model

Risk and return in the CAPM

risk premium and return
return decomposition
diversifiable and systematic risk

Annex: quadratic programming

Risk premium and the Security Market Line
CAPM equation expresses the expected returns µi and µm , as
differences with respect to r , which is the expected return of the
risk-free asset.
This results in the following two definitions.
Definition (risk premium)
Given the expected return of the risk-free asset r , and the expected
return of the i-th risky asset µi , it is said that the difference µi − r is
the i-th risky asset risk premium.

Definition (Security Market Line)

Given the risk premium of a risky asset, the relationship between such
risk premium and the considered risky asset beta is said to be the
Security Market Line (SML) of the risky asset.
Consequently, given an asset’s beta, its expected return can be
determined.
20/31
Next
Introduction

Portfolios with a risk-less asset

Efficient frontier calculation

Capital Asset Pricing Model

Risk and return in the CAPM

risk premium and return
return decomposition
diversifiable and systematic risk

Annex: quadratic programming

Return in the CAPM
The CAPM equation (10) relates the expectations of the returns of
the risky asset Ri , and the market portfolio Rm , with respect to the
riskless asset Rf expected return.
The model can be interpreted, in terms of the random variables Ri ,
Rf , and Rm , as follows

Ri − Rf = αi + βi (Rm − Rf ) + εi , (11)

where αi is a constant and εi is a Gaussian random variable with the

following properties:
i. E [εi ] = 0.
ii. E [Rm , εi ] = 0.
Typically, the variance of εi is denoted by σε2i .
Since by definition Rf is a degenerate random variable, it will be
replaced by its expectation r from now on.
21/31
Return components
The first summand of last equation left hand side gives the following
Definition (alpha)
The parameter αi in the equation (11), modeling the i-th risky asset
return, is said to be the alpha of such risky asset.
A risky asset α reflects the expected return excess over its risk
premium1 .
So assets whose αi ̸= 0 are the ones that would be interesting to find.
This is known as the alpha seek in the asset management industry.
According to the CAPM theory, if the market is in equilibrium, then
αi = 0, and the following two components of the risk premium can be
distinguished:
non
systematic systematic
z }| { z}|{
Ri − r = βi (Rm − r ) + εi .

1
A positive α could indicate a premium for taking other risks. 22/31
Next
Introduction

Portfolios with a risk-less asset

Efficient frontier calculation

Capital Asset Pricing Model

Risk and return in the CAPM

risk premium and return
return decomposition
diversifiable and systematic risk

Annex: quadratic programming

Risk components
As has been done for the expected return, the total variance of Ri can
be decomposed into
total systematic non-system.
variance variance variance
z }| { z }| { z }| {
σi2 = βi2 σm
2
+ σε2i . (12)

Definition (non-systematic risk)

The standard deviation of the random variable εi , denoted by σεi , is
said to be the own risk or non-systematic risk of the risky asset i.
According to the modeling hypotheses of εi the non-systematic risk
can be diversified.
Definition (systematic risk)
The part of the total risk, as decomposed in equation (12), given by
βim σm is called the systematic risk or non-diversifiable risk of the
risky asset i.
23/31
Systematic or non-diversifiable risk
If the risk can be diversified, it implies that it can be arbitrarily
reduced by adding more securities to the portfolio.
Therefore, from the decomposition given by the equation (12), one
relevant conclusion is that the risk that “matters” is the systematic
one.
Since σm is common to all risky assets, then the beta of the asset, βi
is the one that would “measure” the systematic risk.
If the market portfolio reflects properly the “market sentiment”, then
the assets with the largest betas are the most sensitive to market
volatility.
Black (1993) carried out an analysis of the empirical behavior of the
CAPM.
Specifically, he analyzed the relationship between beta and returns on
various equity securities.

24/31
Concluding remarks
Efficient portfolios:
i. They are made up of two assets: the market portfolio; and the
risk-free asset.
ii. Its calculation can be done in two independent steps according to
the Theorem 1 (Tobin separation theorem).
Regarding the CAPM:
iii. It determines the relationship or the trade-off between risk and
expected return in an equilibrium market.
iv. The risk premium is proportional to the beta.
v. It breaks the risk down into systematic, measured by beta, and
non-systematic.
vi. It is difficult to verify empirically, e.g., calculation of the market
portfolio is far from trivial.
vii. It is a very simple model, i.e., single risk factor.
25/31
Next

Introduction

Portfolios with a risk-less asset

Efficient frontier calculation

Capital Asset Pricing Model

Risk and return in the CAPM

Annex: quadratic programming

Maximizing the Sharpe ratio
In this Annex the vectors belong to the vector space Rn := Rn×1 over
the field R. They are denoted in bold lower case letters, and
superscript T means vector transposition.
Matrices are elements of vector space Rn×n over the same field R,
and are denoted by bold capital letters.
The problem of maximizing the Sharpe ratio is given by
µT x − r
maxn √ ,
x∈R xT Σx
s.t.
(13)
1T x = 1,
Ax ≥ b,
x ≥ 0.
Matrix A and vector b are used to include generic restrictions on the
portfolio weights.
26/31
Notation of the maximization problem
In the optimization problem stated by (13), the following notation has
been used
n := number of risky assets in the market.
x := vector with the proportions of each risky asset in the
portfolio.
µ := vector containing the expected return of each risky asset.
r := risk free asset expected return.
Σ := covariance matrix between the returns of each risky asset.
A := matrix of scalars.
b := vector of scalars.
aij := element of A placed at row i and column j.
bk := k element of b.

27/31
Equivalent quadratic program
Proposition 4
Let f : Rn → R be
µT x − r µ̂T x
f (x) := √ =√ ,
xT Σx xT Σx

then for each x such that 1T x = 1, and each scalar λ > 0, it is

verified that f (λx) = f (x).

Proof
Taking y = λx, following equalities are obtained

µ̂T y = λµ̂T x,
q √
yT Σy = λ xT Σx,

their ratio proves the statement of the proposition. ■

28/31
Equivalent quadratic program (ii)
Proposition 5
The optimization program given by (13) is equivalent to

1
maxn p ,
y∈R yT Σy
s.t.
(14)
µ̂T y = 1,
Ây ≥ 0,
y ≥ 0.

where
µ̂ := µ − r 1
 denotes the matrix whose element in row j, and in column k is
given by âjk := ajk − bj .

29/31
Equivalent quadratic program (iii)
Proof
Let y∗ be a feasible point of (14), and be

y∗
x∗ := .
1T y∗

First it is proved that x∗ is also a feasible solution of (13).

Positivity is trivially fulfilled, and by definition 1T x∗ = 1.
On the other hand, Ây∗ ≥ 0 implies that for any row i of  it is
verified
n
(aij − bi ) yj∗ ≥ 0,
X

j=1

this is
n  
aij yj∗ ≥ 1T y∗ bi .
X

j=1

30/31
Equivalent quadratic program (iv)
Proof (continued)
Consequently
n
aij xj∗ ≥ bi ,
X

j=1

and therefore it is verified that Ax∗ ≥ b.

For all this it can be concluded that if y∗ a feasible solution of (14)
then x∗ is too a feasible solution of (13).
Finally, since µ̂T y∗ = 1, and according to Proposition 4

f (x∗ ) = f (y∗ )
1
=p .
yT Σy

If y∗ is a point with zero gradient, then x∗ is too.

The reverse implication is proven similarly. ■
31/31
References I
Black, Fisher (1972). “Capital Market Equilibrium with
Restricted Borrowing”. In: Journal of Business 45, pp. 444–455.
— (1993). “Beta and Return”. In: The Journal of Portfolio
Management 20, pp. 8–18.
Lintner, John (1965a). “Security prices, risk, and maximal gains
from diversification”. In: Journal of Finance 20, pp. 587–615.
— (1965b). “The Valuation of Risk assets and the selection of
Risky Investments in Stock Portfolios and Capital Budgets”. In:
Review of Economics and Statistics 47, pp. 13–37.
Mossin, Jan (Oct. 1966). “Equilibrium in a capital asset
market”. In: Econometrica 34, pp. 768–783.
References II
Sharpe, William (1964). “Capital asset prices: A theory of
market equilibrium under conditions of risk”. In: Journal of
Finance 19, pp. 425–442.
— (1970). Portfolio Theory and Capital Markets. Ed. by
McGraw-Hill Education. McGraw-Hill.
Tobin, James (1958). “Liquidity Preference as Behavior Towards
Risk”. In: Review of Economic Studies 25, pp. 65–85.