Scribd
Upload
9 views

Lecture 9

Lecture

Uploaded by

loydolagodamontanha
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
9 views

Lecture 9

Lecture

Uploaded by

loydolagodamontanha
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 22

CAPM Review, Factor Models and

Arbitrage Pricing Theory

Lecture 9
CAPM Review
z The Market Portfolio is the efficient portfolio.
z Systematic risk is rewarded with higher returns,
idiosyncratic risk isn’t.

( )
z The SML:
E [ ri ] = rf + βi E ⎡⎣ rM − rf ⎤⎦
where :
σ iM
βi = 2
σM

z Uses of CAPM: Pricing of securities, to calculate


discount factors for investment projects, regulators.
z Derivation of the CAPM
Example: GM
Security Characteristic
Line for GM: Summary Output
Hands On Beta Calculation
z Choose 5 stocks.
z Go to finance.yahoo.com or other financial website.
z Download weakly historical prices for the 5 stocks of your
choice. Historical prices for the S&P 500 and historical treasury
bill rates.
http://www.federalreserve.gov/releases/h15/data/Weekly_Wednesday_/H15_FF_O.txt
z Calculate the monthly excess returns for the stock and the
excess return of the market. (take into account dividends)
z Regress the Stock returns against a constant and the excess
return of the market. What are the alphas and betas?
z Redo the regression but break the sample into two sub-samples.
Are your estimates of betas and alphas the same? How does
your estimate of Beta compare to the Yahoo or other available
estimate?
The CAPM is Linear
The price of a sum of assets is the sum of their prices. The
price of the multiple of an asset is that same multiple of
the basic price. P = Q
P =
Q 1 2

1 + rf + β1 ( rm − rf ) 1 + rf + β 2 ( rm − rf )
1 2

Q1 Q2 Q12
P1 + P2 = + =
1 + rf + β1 ( rm − rf ) 1 + rf + β 2 ( rm − rf ) 1 + rf + β12 ( rm − rf )

Where : Q12 = Q1 + Q2 and β12 = beta of the combined asset.[ Hmk ]


Alternatively :
⎡Q ⎤
Cov ⎢ i − 1, rm ⎥
βi = ⎣ Pi ⎦ = Cov [Qi , rm ] => P = Qi
=
Qi
σ m2 Piσ m2 1 + rf + βi ( rm − rf ) Cov [Qi , rm ] ( rm − rf )
i

1 + rf +
Piσ m2
Cov [Qi , rm ] ( rm − rf ) =Q ⎡ Cov [Qi , rm ] ( rm − rf ) ⎤⎥
Pi (1 + rf ) +
1
=> Pi = ⎢Qi −
σ m2 i
(1 + rf ) ⎢⎣ σ m2 ⎥⎦

1 ⎡ Cov [Q1+ 2 , rm ] ( rm − rf ) ⎤⎥
P1+ 2 = ⎢Q1+ 2 −
(1 + rf ) ⎢⎣ σ m2 ⎥⎦
The contribution of a stock
to the volatility of a portfolio

The volatility of a portfolio with $1 in the Market


Portfolio and $ε in portolio i :
σ p ( ε ) = (σ + ε σ + 2 × 1× εσ iM )
2 2 2 1/ 2
M i

dσ p ( ε ) 1 2εσ i2 + 2σ iM
=
dε 2 (σ 2 + ε 2σ 2 + 2 × 1× εσ )1/ 2
M i iM

dσ p ( 0 ) σ iM σ iM σ iM σ M
= = = = βiM σ M
dε (σ )2 1/ 2
M
σM σM σM
Discussion about CAPM
The evidence against the CAPM can be summarized as follows:

• First, for some sample periods, the relation between average return
and beta is completely flat.
• Second, other explanatory variables such as firm size (market
equity) and the ratio of book-to-market equity seem to do better than
beta in explaining cross-sectional variation in average asset returns.
[Fama and French]

z Practical: When the universe of assets is very large, calculating all the
information needed for determining the optimal weights in the mean-
variance approach becomes unpractical. For n assets a total of 2n+n(n-
1)/2 parameters need to be estimated.

z Equity Premium (Excess Return) Puzzle: Hard to justify that people are
so risk averse that historically the return on the market portfolio exceeds
the return on the risk free security by about 7%. (This puzzle might have
been significantly reduced this last year!!)
Factor Models
z Practical Motivation:
— Difficulty of calculating the parameters needed to
do Markowitz (for n=1000 there are 500,000 +
parameters that need to be estimated). Factor
models seek to uncover some underlying sources of
randomness (factors) that affect the returns of all
stocks. This greatly reduces the amount of
parameters that need to be estimated.
— Taking the excess return of the Market portfolio
as one of the factors we can see if there are other
factors that help predict returns. (Theory vs.
Data Mining)
Fama French Research

Returns are related to factors other than market


returns
z Size (Market Capitalization)
z Book value relative to market value
z Three factor model better describes returns
The Size Effect
from 1926 to 2003
Average Rate of Return
as a Function of Book to Market
Factor Models Example
Arbitrage Pricing Theory
z Assumptions:
— Returns are generated by a linear factor model.
— There are very many (∞) assets.
— There are NO arbitrage opportunities.
z Result:
— The expected excess return of a portfolio is given
by the sum of the exposure of the portfolio to
different risk factors times the market price of
those factors.
z For Example if the factor is the excess return of the
market then we get a CAPM representation for
portfolios. (more on this later).
Factor Models Motivation
z Practical: When the universe of assets is very
large, calculating all the information needed
for determining the optimal weights in the
mean-variance approach becomes unpractical.
For n assets a total of 2n+n(n-1)/2
parameters need to be estimated.
z More systematic risk factors in addition to
the “market” portfolio.
A Linear Factor Model
Analogy with the CAPM
z Let K=1 and the only factor be the excess
return on the market portfolio.
μi = ρ + βi λr m

E [ ri ] = ρ + β i E ⎡⎣ rm − rf ⎤⎦
⎛ ρ must equal rf as in ⎞
E [ ri ] = rf + β i E ⎡⎣ rm − rf ⎤⎦ ⎜ ⎟
⎝ CAPM otherwise AO ⎠
CAPM implies an exact relationship stock by stock APT holds
almost surely stock by stock (it is designed for large portfolios)
APT and CAPM Compared

z APT applies to well diversified portfolios and not


necessarily to individual stocks
z With APT it is possible for some individual stocks to be
mispriced - not lie on the SML
z APT is more general in that it gets to an expected
return and beta relationship without the assumption of
the market portfolio
z APT can be extended to multifactor models
APT Proof (no idiosyncratic risk)
Suppose that there are n assets whose rates of return
are governed by K<n factors according to the equation
k
ri = ai + ∑ bij f j for i = 1,..., n
j =1

Then there are constants λ0 , λ1 ,..., λK such that :


k
ri = λ0 + ∑ bij λ j for i = 1,..., n
j =1

Strategy of proof:
z First we construct zero beta portfolios and argue by
arbitrage.
z Next we use linear algebra and argue by
contradiction.
APT Proof (no idiosyncratic risk)
Choose ω such that ω ′β k = 0 for k = 1, 2..., K and ω ′1 = 0
This portfolio has zero cost and zero risk .
By No Arbitrage => ω ′ri = 0
k
Suppose ri ≠ λ0 + ∑ bij λ j for some i = 1,..., n
j =1

Let l j j = 1,..., K be the best linear fit of r onto ⎡⎣1, b j ⎤⎦


k
r = l01 + ∑ bij l j + ei
j =1

Choosing l optimaly implies:


e′1 = 0 and e′b j = 0 for j = 1,..., K
Nowlet e be the weights of a costless and riskless portfolio with return :
k
e′r = e′a + ∑ ⎡⎣e′bij ⎤⎦ f j
j =1
k
e′r = e′1l0 + ∑ ⎡⎣e′bij ⎤⎦ l j + e′e = e′e > 0 !!! An arbitrage oportunity !!!
j =1

You might also like