Anda di halaman 1dari 17

Single Index Model

A. Single Index Model

The CAPM is a theory about expected returns


The application of the CAPM, i.e., the empirical
version, is ex-post, or after the fact
The empirical version is often referred to as the
Single Index Model

One step removed from the theoretical CAPM and all of


its assumptions

Single Index Model

A broad stock market index is assumed to be the


single, common factor for all stocks

(ri rf ) i i (rm rf ) ei
i = expected return of stock i if markets excess return is zero
i(rmt - rft) = component of return due to market movements
eit = component of return due to unexpected firm-specific events

Single Index Model

Textbook notation:
Ri = ri rf and Rm = rm - rf
Therefore,

Ri i i Rm ei

Early Application

To simplify the Markowitz model


Inputs of the Markowitz model: means, standard
deviations, and covariances (or correlation
coefficients) of the assets
If you have 25 assets in the investment universe
how many unique covariances?
n(n-1) 2
= 300

Simplifying the Markowitz Model

Adopting the Single Index Model is a way to reduce


this number

According to the model,

By simplifying the covariance


All asset returns derive only from the common factor, RM
ei is firm-specific, and hence uncorrelated across assets

Therefore,
Cov(Ri, Rj) = Cov(iRM, jRM ) = ijs2M

Implication for Security Analysis

This setup allows security analysts to specialize

Provides rationale for why analysts do not have to


research other sectors
Model says only the common factor (the market) matters;
there is no relationship otherwise

Decomposing Total Risk

Single Index Model for a portfolio of stocks:

p p Rm e p
The variance ofRR
p pis:

As the number of2stocks


increases,
the last term
2 2
2
s p psas
(e p )of diversification
m
becomes less important
asresult
Total risk = systematic risk + diversifiable risk

If Portfolios are equally weighted...

Pink curve: total risk. Can exclude proof on


pp.276-7

Estimating the Single Index Model

Regression analysis

Rit i i Rmt eit

Typically, use monthly returns over the past 5 years


(i.e., 60 observations) to estimate
Y: excess return on individual security (or individual
portfolio)
X: excess return on market index

Intercept is i, slope is

Security Characteristic Line

Interpreting the Results

alpha
beta

statistical significance

The Meaning of R2
The goodness-of-fit measure, R2, from the
Single Index Model regression (the SCL) is:
2 2
2

s
s
(ei )
2
i
m
R
1
2
si
s i2

In words, the R2 = the percentage of total risk


of asset i that can be explained by its
systematic risk

Industry Versions

BMO Nesbitt Burns, Merrill Lynch, Value Line

These (and several other) beta estimate providers use


raw returns, not excess returns
That model is called the Market Model

Some firms forecast beta as a function of past betas


Some firms forecast beta as a function of firm size,
growth, leverage, etc.

Industry Versions

Bank of America Merrill Lynch

Adjusted : 2/3 sample beta and 1/3 beta of one


Adjusted = 2/3 + 1/3
Tendency for to move toward one over time Hence, take
this into account in forecasts

Beta books

Merrill Lynch: monthly


Ibbotson Associates: semi-annual

Market Neutral Strategies

An application of the Single Index Model


A long/short market neutral investment strategy:
Extract the alpha of another managers portfolio
Example on p.288:

To extract alpha, need to get rid of the exposure to


the TSX

R p 0.04 1.4 RTSX e p

Market Neutral Strategies

First, define the following tracking portfolio, T:

1.4 RTSX

Can think of T as a leveraged portfolio: 1.4 in the TSX,


and -0.4 in risk-free asset
Typo in text: share in risk-free asset should be -0.4,
not 20.4
End result:

Rc R p RT 0.04 e p

Anda mungkin juga menyukai