INVESTMENTS | BODIE, KANE, MARCUS
Copyright © 2011 by The McGraw-Hill Companies, Inc. All rights reserved. McGraw-Hill/Irwin
CHAPTER 8
Index Models
INVESTMENTS | BODIE, KANE, MARCUS
Chapter Overview
• Advantages of a single-factor model
• Risk decomposition – Systematic vs. firm-specific
• Single-index model and its estimation
• Optimal risky portfolio in the index model
– Index model vs. Markowitz procedure
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Advantages of the Single Index Model
• Reduces the number of inputs for diversification
• Easier for security analysts to specialize
8-3
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Single Factor Model
r𝑖 = 𝐸[𝑟𝑖] + 𝛽𝑖𝑚 + 𝑒𝑖 Where:
𝛽𝑖 response of an individual security’s return to the common factor m. Beta measures systematic risk.
𝑚 a common macroeconomic factor that affects all security returns. The S&P500 is often used as a proxy for 𝑚
𝑒𝑖 firm-specific surprises
8-4
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Single-Index Model
Regression Equation:
𝑅𝑖 𝑡 = 𝛼𝑖 + 𝛽𝑖𝑅𝑀 𝑡 + 𝑒𝑖 𝑡
The expectation of the residual term 𝑒𝑖 is zero, so the expected return-beta relationship is:
𝐸 𝑅𝑖 = 𝛼𝑖 + 𝛽𝑖𝐸 𝑅𝑀
8-5
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Single-Index Model
Risk and covariance:
• Variance - Systemic risk and firm-specific risk, assume noise is uncorrelated:
• Covariance - product of betas x market index risk:
8-6
2 2 2 2 ( )i i M ie
2( , )i j i j MCov r r
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Single-Index Model - Correlation
Product of correlations with the market index:
8-7
ji
Mji
jiji rrCorr
2
, ,
MjMi
MjMi
22
MjMi rrCorrrrCorr ,,
MM
M
2
jijijijiji rrCovrrCorr /,, ,,
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Questions to test your intuition
• What is the stock’s E[𝑟] if (𝑟𝑀−𝑟𝑓) = 0 ?
• What is the responsiveness of the stock to market movements relative to 𝑟𝑓?
• What is the stock-specific component of return (not driven by the market)?
• What is the variance attributable to uncertainty of the market?
• And that attributable to firm-specific events?
8-8
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Index Model and Diversification
• Consider an Equally weighted portfolio and take the expected return 𝑅𝑃 as the average:
8-9
n
i
iMii
n
i
iP eRn
Rn
R11
11
n
i
in 1
1
n
i
MiRn 1
1
n
i
ien 1
1
PPR MPR Pe
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Index Model and Diversification
• The portfolio variance by definition: 𝜎𝑃2 = 𝛽𝑃
2𝜎𝑀𝑃 + 𝜎2(𝑒𝑃)
• where the market component comes from the portfolio’s sensitivity to the market:
• and the non-systemic component 𝜎2(𝑒𝑃) is the contribution of all the stocks in the portfolio.
8-10
n
i
iPn 1
1
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Index Model and Diversification
• Variance of the non-systemic component of an equally weighted portfolio is (we assume all the stock-specific components are uncorrelated):
𝜎2(𝑒𝑃) = 1
𝑛
2
𝜎2(𝑒𝑖)
𝑛
1
=1
𝑛𝜎 2(𝑒)
• When n gets large, 𝜎2(𝑒𝑃) becomes negligible and firm specific risk can be diversified away.
8-11
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Figure 8.1 The Variance of an Equally Weighted Portfolio with Risk Coefficient βp
8-12
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Fig 8.2 Excess Returns on HP and S&P500
8-13
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Fig 8.3 Scatter Diagram of HP, the S&P 500, and HP’s Security Characteristic Line (SCL)
8-14
tetRtR HPSPHPHPHP 500
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Table 8.1 Excel Output: Regression Statistics for the SCL of Hewlett-Packard
8-15
(monthly)
correlation
explanatory power
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Table 8.1 Interpretation
• Correlation of HP with the S&P500 is 0.7238
• The model explains about 52% of the variation in HP
• HP’s alpha is 0.86% per month (10.32% pa), but it is not statistically significant
Q. What does it mean? Why?
• HP’s beta is 2.0348, but the 95% confidence interval (which is +/- ~2 standard errors) is quite wide (~2 x 0.25 = 0.5)
8-16
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Figure 8.4 Excess Returns on Portfolio Assets
8-17
• Study pairs of securities
vs the market to
estimate correlations
• Compute stats to
measure correlations
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Study portfolio stats – 1
8-18
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
A closer look at correlations
8-19
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Study portfolio stats – 2
8-20
2
Mji
iM ei
222
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Example: build optimal portfolio
8-21
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Alpha and Security Analysis
1. Use Macroeconomic analysis to estimate risk premium and risk of the market index (𝑅𝑀 , 𝜎𝑀)
2. Use statistical analysis to estimate the beta coefficients of all securities and their residual variances 𝜎2(𝑒𝑖)
8-22
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Alpha and Security Analysis
3. Use numerical methods to establish the expected return of each security independently of security analysis (𝛽)
4. Use security analysis to develop your own forecast of the expected returns for each security (𝛼)
8-23
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Single-Index Model considerations
• Techniques for estimating 𝛽 are well known
• Estimating alpha requires a deep knowledge of the company behind the stock:
– Positive 𝛼 means overweight in the
portfolio
– What do you do if 𝛼 is negative?
8-24
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Recall the Minimum-Variance Frontier
7-25
Chapter 7
took the
entire
universe of
stocks and
used
brute-force
math to
find the
efficient
frontier
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Single-Index Model – Optimization
• Single-Index model offers a simpler optimization than the model in chapter 7 as the model is simplified
• Include the market as asset n+1 to improve diversification. By definition:
– Beta of market index = 1
– Alpha of market index = 0
– emarket_index = 0
8-26
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Single-Index Model Input List
• Risk premium on the S&P500 portfolio (𝑅𝑀)
• Estimate of the SD of the S&P500 portfolio (𝜎𝑀)
• n sets of estimates (one set for each stock) of:
–Beta coefficient
–Stock residual variances
–Alpha values
8-27
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Single-Index Model steps
• Use RM, alphas and betas to construct n+1 expected returns
• Use betas and 𝜎𝑀 to construct the covariance matrix
• Set up the optimization problem to minimize portfolio variance, given a return, subject to…
• …constraint that weights add up to one
• You could use excel solver to solve this problem and build your efficient frontier
8-28
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Index Model – Recall 𝛼𝑃 and 𝛽𝑃
Consider a generic portfolio and take the excess return RP as the average:
8-29
n
i
iMiii
n
i
iiP eRwRwR11
n
i
iiw1
n
i
Mii Rw1
n
i
iiew1
PPR MPR Pe
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Optimal Risky Portfolio of the Single-Index Model
Now take the portfolio expected excess return:
8-30
PMPP RERE
n
i
iiM
n
i
ii wREw11
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Optimal Risky Portfolio of the Single-Index Model
Standard Deviation and Sharpe Ratio:
8-31
PMPP e2222
n
i
iiM
n
i
ii eww1
222
2
1
PPP RES /
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Optimal Risky Portfolio of the Single-Index Model
• No need to use Excel as there is an analytical solution
• Solution is a combination of:
–Active portfolio (A), with weight wA
–Market-index passive portfolio (M)
8-32
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Optimal Risky Portfolio - wA
Assume for a moment beta=1
Then the optimal weight wA is proportional to
the ratio 𝜎𝐴/𝜎2(𝑒𝐴) to balance excess return
and residual variance from Active portfolio A:
8-33
2
20
M
M
A
A
A RE
ew
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Optimal Risky Portfolio of the Single-Index Model
Next, modify of active portfolio weight wA to
optimize, as beta is not necessarily =1:
8-34
Notice that when
0
0*
11 AA
AA
w
ww
0* then 1 AAA ww
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
The Information Ratio
The Sharpe ratio of an optimally constructed risky portfolio will exceed that of the index portfolio (the passive strategy):
𝑆𝑃2 = 𝑆𝑀
2 +𝛼𝐴𝜎 𝑒𝐴
2
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
8-35
Information
Ratio
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Figure 8.5 Efficient Frontiers with the Index Model and Full-Covariance Matrix
8-36
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Table 8.2 Portfolios from the Single-Index and Full-Covariance Models
1-37
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
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 decouples macro and security analysis.
8-38
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS 8-39
Beta Book: Industry Version of the Index Model
• Use 60 most recent months of price data
• Use S&P500 as proxy for M
• Compute total returns that ignore dividends
• Estimate index model without excess
returns: *ebrar m
instead of
errrr fmf
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS 8-40
Beta book – alpha and intercept
• The intercept is different in the two
formulas. Rewrite as:
If* rf is constant you have same 𝛽 and 𝑒.
The intercept a is an estimate for
errrr fmf
errr mf 1
1fr
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Beta Book: Industry Version of the Index Model
• The average beta over all securities is 1. Thus, our best forecast of the beta would be that it is 1.
• Also, firms may become more “typical” as they age, causing their betas to approach 1.
Adjust beta because:
8-41
13
13
2 sampledadjusted
RESET STYLE
INVESTMENTS | BODIE, KANE, MARCUS
Table 8.4 Industry Betas and Adjustment Factors
1-42
RESET STYLE