CHAPTER 8

Index Models
 Index Models
• The Markowitz model requires a long list of estimates of
expected security returns and the covariance matrix. (p.
257)

8-2
 Index Models (Contd.)
• The estimated covariance/correlation matrix may not
always be consistent (p. 257)

The single index model posits that covariances among assets are due to
the influence of the single common factor represented by the market.

8-3
 Single-factor Model
• The rate of return on security-i can be shown as the sum
of an expected and an unexpected components
ri  E (ri )   i
• The single factor model posits that the uncertain
component may be influenced by a common factor m
and a firm-specific factor, which are independent of each
other

•
ri  E ( ri )   i m  e
reflects sensitivity to the macro factor.
i Note, for
example, pharmaceuticals firms may not respond as
much as auto firms to economic downturns.

8-4
 Single-factor Model
• Therefore, the asset return variance is due to
uncertainties associated with the common factor
(systematic risk) and firm-specific factor (unsystematic
risk)
 i2   i2 m2   2 (ei )
Total Risk = Systematic Risk + Firm-specific Risk

• The common and firm-specific factors being unrelated,

the covariance of any pair of assets is governed by the
respective asset’s sensitivity to the common factor

Cov(ri , rj )  Cov( i m  ei ,  j m  e j )  i  j m2
8-5
 Single-factor Model

• Correlation between a pair of assets is therefore

Cov(ri , rj ) i  j m2
Corr (ri , rj )  
 i j  i j
i  j m2  m2 i m2  j m2 i m2  j m2
  2   
 i j  m  i m j m  i m  j m
 Corr (ri , rm )  Corr (rj , rm )

8-6
 Single-Index Model
• The most commonly used empirical specification of the
single factor (market index) model (where is excess
return):
Ri  t    i   i Rm  t   ei  t 
• As = 0, expected return-beta relationship:
E  Ri    i   i E  Rm 
Security Characteristic Line (SCL)
• Therefore, the parameter estimates needed under the
index model are
2
 i ,  i ,  (ei ), E ( Rm ), and  ! m

8-7
 Single-Index Model

• Note that the index model does not require pair-wise

covariance analysis. Therefore, a less obvious advantage
of the index model abstraction is that it encourages
specialization in the security analysis industry. You don’t
need to have a deep understanding of all industries,
which would be necessary to make an informed
judgment of co-movement among industries under the
Markowitz framework.

8-8
 Single-Index Model

• A disadvantage is that it imposes a rather restrictive

structure on the structure of asset return uncertainty –
market vs. firm-specific risk. It rules out the role of other
possible sources of real-world uncertainty. For example,
industry events may affect many firms within an industry
without substantially affecting the broad macroeconomy
(p. 262 for details).

8-9
 Index Model and Diversification
• Assuming an equally weighted portfolio = we
get
• =∑ = ∑

= ∑ ( + + )

= ∑ +( ∑ ) + ∑
• Therefore, = + + and
= + ( )

8-10
Index Model and Diversification
• Variance of the equally weighted portfolio of
firm-specific components:
n 2
2 1 2 1 2
 (eP )      (ei )   (e)
i 1  n  n

• When n gets large, σ2(ep) becomes negligible and

firm specific risk is diversified away.

8-11
Figure 8.1 The Variance of an Equally Weighted Portfolio
with Risk Coefficient βp

8-12
Figure 8.2 Excess Returns on HP and S&P 500

8-13
Figure 8.3 Scatter Diagram of HP, the S&P 500, and HP’s
Security Characteristic Line (SCL)

RHP t    HP   HP RS & P 500 t   eHP t 

8-14
Table 8.1 Excel Output: Regression Statistics for
the SCL of Hewlett-Packard

8-15
 Table 8.1 Interpretation

• Correlation of HP with the S&P 500 is 0.7238.

• The model explains about 52% of the variation in HP.
• HP’s alpha is 0.86% per month(10.32% annually) but
it is not statistically significant.
• HP’s beta is 2.0348, but the 95% confidence interval
is 1.43 to 2.53.

8-16
 Table 8.1 Interpretation

• Sum of squares (SS) of regression (0.3752) is the proportion of

variance HP as explained by S&P500, which is equal to
2 2
 
• Mean sum of squares (MS) for the residual (0.0059) is the
HP S & P 500

variance of the unexplained portion of HP returns , the

square root of which is the standard error of regression 2 (ei )
(0.0767).
• Total SS of regression (0.7162) divided by 59 is the total
variance of HP return.
• Dividing explained SS of regression (0.3752) by total SS (0.7162)
gives R-square of regression.

8-17
Figure 8.4 Excess Returns on Portfolio Assets

8-18
 Alpha and Security Analysis
1. Use macroeconomic analysis to estimate the risk
premium and risk of the market index.

2. Use statistical analysis to estimate the beta

coefficients of all securities and their residual
variances, σ2 (ei).

8-19
 Alpha and Security Analysis

3. Establish the expected return of each security

absent any contribution from security analysis.

4. Use security analysis to develop private forecasts

of the expected returns for each security.

8-20
 Single-Index Model Input List

• Risk premium on the S&P 500 portfolio

• Estimate of the SD of the S&P 500 portfolio
• n sets of estimates of
– Beta coefficient
– Stock residual variances
– Alpha values

8-21
 Optimal Risky Portfolio of the
Single-Index Model
• Maximize the Sharpe ratio
– Expected return, SD, and Sharpe ratio:

n 1 n 1
E ( RP )   P  E ( RM )  P   wi i  E ( RM ) wi  i
i 1 i 1
1
2
2 2 2
1  2  n 1
 n 1
2 2
 2
 P    P M   (eP )    M   wi i    wi  (ei ) 
2

  i 1  i 1 
E ( RP )
SP 
P

8-22
 Optimal Risky Portfolio of the
Single-Index Model
• Combination of:
– Active portfolio denoted by A

– Market-index portfolio, the passive portfolio

denoted by M

8-23
 Optimal Risky Portfolio of the
Single-Index Model
Modification of active portfolio position:
0
* w A
wA 
1  (1   A ) wA0

When

* 0
 A  1, w  w
A A

8-24
 The Information Ratio
• The Sharpe ratio of an optimally
constructed risky portfolio will exceed that
of the index portfolio (the passive strategy):

2
2 2  A 
s P  s M    (eA ) 

8-25
 The Information Ratio
• The contribution of the active portfolio
depends on the ratio of its alpha to its residual
standard deviation.

• The information ratio measures the extra

return we can obtain from security analysis.

8-26
Figure 8.5 Efficient Frontiers with the Index
Model and Full-Covariance Matrix

8-27
Table 8.2 Portfolios from the Single-Index and
Full-Covariance Models

8-28
 Is the Index Model Inferior to the
Full-Covariance Model?
• Full Markowitz model may be better in principle, but

– Using the full-covariance matrix invokes

estimation risk of thousands of terms.
– Cumulative errors may result in a portfolio that is
actually inferior to that derived from the single-
index model.
– The single-index model is practical and
decentralizes macro and security analysis.

8-29
 Beta Book: Industry Version of the Index
Model

• Use 60 most recent months of price data

• Use S&P 500 as proxy for M
• Compute total returns that ignore dividends
• Estimate index model without excess returns:

*
r  a  brm  e

8-30
 Beta Book: Industry Version of the Index
Model

• The average beta over all

securities is 1. Thus, our
best forecast of the beta
Adjust beta would be that it is 1.
because:
• Also, firms may become
more “typical” as they age,
causing their betas to
approach 1.
8-31
Table 8.4 Industry Betas and Adjustment
Factors

8-32