Index Model

CHAPTER 4
DR. LOUKIA EVRIPIDOU

CFS 563 1
Drawbacks of Markowitz Portfolio Theory
•Computational Complexity – Requires calculating and storing an N×N covariance
matrix for N assets, making it impractical for large portfolios.
•Estimation Errors – Small errors in expected returns, variances, or covariances can
significantly affect portfolio allocation.
•Assumption of Normality – Assumes asset returns are normally distributed, which
isn't always true (e.g., fat tails, skewness).
•Stationarity Assumption – Assumes past data reliably predicts future returns and
correlations, which may not hold in dynamic markets.
•Ignoring Market Factors – Doesn't account for macroeconomic or industry-wide
factors affecting asset returns.
•Liquidity and Transaction Costs – Ignores real-world frictions such as transaction
costs, liquidity constraints, and taxation.

CFS 563 2
A Single-Factor Security Market

The Single Index Model (SIM) is a simplified approach to portfolio theory that assumes the
returns of individual assets are driven primarily by a single common factor—typically a broad
market index (e.g., S&P 500). It was developed by William Sharpe to reduce the complexity of the
Markowitz Portfolio Theory while still providing a useful framework for portfolio selection and
risk analysis.
Advantages of the single-index model.
1. Simplifies Computation: Reduces the need to calculate the full covariance matrix in
Markowitz's model.
2. Efficient Estimation: Requires estimating only beta and residual variances, rather than all
pairwise asset correlations.
3. Practical for Large Portfolios: Easier to implement when managing portfolios with many
assets.

3
Single-Index Model -Regression equation.

Ri (t )  α i  β i RM (t )  ei (t )

4
Figure 8.1 Scatter diagram for U.S. Steel against the market index and U.S. Steel’s
security characteristic line, 5 years ending December 2021.

5
Expected return–beta relationship.

E ( Ri )  αi  βi E ( RM )
m  Market factor that measuring unanticipated macroeconomic developments
βi  Sensitivity coefficient for firm i ei  Firm-specific random variable

6
Risk and Covariance in the single index model

Total risk = Systematic risk + Firm-specific risk.

σi2  β i2 σ 2M  σ 2 (ei )

Covariance = Product of betas × Market-index risk.

Cov  ri , rj   βi β j σ 2M

Correlation = Product of correlations with the market index.

7
8
CFS 563 9
The Set of Estimates Needed for the Single-Index Model

Access the text alternative for slide images.

10
Index Model and Diversification 2

Variance of the equally weighted portfolio of firm-specific components:

2
 1 n
 n
 1  1 n
σ 2
(ei ) 1 2
σ 2 (eP )  Var   ei      σ 2 (ei )    σ ( e)
 n i 1  i 1  n  n i 1 n n

When n gets large,  2 (e p ) becomes negligible.

As diversification increases, the total variance of a portfolio approaches the

systematic variance.

11
12
Index Model and Diversification

Portfolio Variance is:

σ 2P  β 2P σ 2M  σ 2 (e p )

β 2P σ 2M  Systematic risk of the portfolio

(depends on average beta of the individual securities)

σ 2 (eP )  Firm-specific risk of the portflio

13
Figure 8.3 The Variance of an Equally Weighted Portfolio
With Risk Coefficient, βp

Access the text alternative for slide images.

14
 Security Characteristic Line (SCL)

The SCL describes

the (linear)
dependence of a
firm’s excess return
on the excess
return of the
market index
portfolio, in this
case represented
by the S&P500. It’s
important to note
that other indices
may be used
instead of the
S&P500.

15
Figure 8.3 Excess Monthly Returns on Amazon and the Market Index

Figure 8.3 Excess monthly returns on U.S. Steel and the market index, 5 years ending
December 2021.

Access the text alternative for slide images.

16
Table 8.1 Excel Output: Regression Statistics
Regression Statistics
Multiple R 0.532
R2 0.283
Adjusted R2 0.271
Standard Error 15.171
Observations 60

Explanatory variable Coefficient Standard Error t–Statistic p–Value

Intercept −2.034 2.052 −0.991 0.326
Market index 2.030 0.424 4.785 0.000

Excel Output: Regression statistics for U.S. Steel’s SCL. Observations are
monthly returns measured as percents (not decimals) from January 2017
through December 2021.

17
18
19
Table 8.2 Index Model Regressions 1

Residual Standard Standard Adjusted

Company Beta Alpha R-Square Std Dev Error Beta Error Alpha Beta
Newmont Mining 0.217 0.011 0.016 0.081 0.227 0.011 0.478
Campbell Soup 0.461 −0.008 0.109 0.062 0.173 0.008 0.640
Intel 0.510 0.002 0.107 0.069 0.193 0.009 0.673
McDonald's 0.528 0.008 0.246 0.043 0.121 0.006 0.685
Pfizer 0.615 0.007 0.176 0.062 0.174 0.008 0.743
Coca-Cola 0.644 0.000 0.319 0.044 0.124 0.006 0.762
Starbucks 0.759 0.004 0.291 0.056 0.156 0.008 0.839
Microsoft 0.806 0.019 0.517 0.037 0.102 0.005 0.871
Alphabet (Google) 1.003 0.009 0.523 0.045 0.126 0.006 1.002
Amazon 1.053 0.012 0.372 0.064 0.180 0.009 1.035
Union Pacific 1.126 0.002 0.587 0.044 0.124 0.006 1.084
Walt Disney 1.175 −0.007 0.459 0.060 0.168 0.008 1.117
ExxonMobil 1.295 −0.018 0.491 0.062 0.173 0.008 1.197
Bonk of America 1.442 −0.005 0.662 0.048 0.135 0.007 1.295

20
21
Table 8.2 Index Model Regressions 2

Residual Standard Standard Adjusted

Company Beta Alpha R-Square Std Dev Error Beta Error Alpha Beta
Boeing 1.474 −0.009 0.306 0.104 0.291 0.014 1.316
U.S. Steel 2.030 −0.020 0.283 0.152 0.424 0.021 1.687
Marathon Oil 2.986 −0.024 0.477 0.147 0.410 0.020 2.324
AVERAGE 1.066 −0.001 0.350 0.069 0.194 0.009 1.044
STD DEVIATION 0.672 0.012 0.184 0.034 0.095 0.005 0.448

Table 8.2
Index model estimate: Regressions of excess stock returns on the excess return of the broad market index over 60 months, 5 years
ending December 2021.

22
The Industry Version of the Index Model: Predicting Betas 1

The Industry Version of the Single Index Model (SIM) refines beta estimation by assuming
that firms within the same industry share similar risk characteristics. Instead of estimating
individual stock betas based only on past market returns, this approach groups companies into
industry sectors to predict their future betas more accurately

23
Why Use the Industry Version?
More Reliable Beta Estimates – Individual firm betas can be noisy due to short-
term fluctuations. Industry betas provide a more stable estimate.
Better Predictions for New or Low-Data Firms – For firms with limited trading
history (e.g., newly listed companies), using the industry beta helps estimate
their systematic risk.
Reduces Random Estimation Errors – Industry betas are based on an aggregate
of similar firms, reducing idiosyncratic noise from a single stock's beta estimate.

24
Industry specific beta - formula
Betas tend to mean-revert over time, making past values imperfect predictors.
Blume (1971) observed that betas tend to move toward the market average (β=1)
over time. His adjustment formula is:
2 1
Adjusted beta =  Estimated beta   1.00
3 3

•Stocks with high betas (>1) tend to decrease, while low betas (<1) tend to
increase.
•Works well for long-term forecasting.

CFS 563 25
The Industry Version of the Index Model: Predicting Betas
Forecasting betas.
Current beta = a  b   Past beta  
Forecast beta = a  b   Current beta 
or
Current beta = a  b1   Past beta   b2   Firm size   b3   Debt ratio  
Forecast beta = a  b1   Current beta   b2   Firm size   b3   Debt ratio 
Rosenberg and Guy found the following variables predictive:
1. Variance of earnings.
2. Variance of cash flow.
3. Growth in earnings per share.
4. Market capitalization (firm size).
5. Dividend yield.
6. Debt-to-asset ratio.
26
Table 8.3 Industry Betas and Adjustment Factors

Industry Beta Adjustment Factor

Agriculture 0.99 −0.140
Drugs and medicine 1.14 −0.099
Energy utilities 0.60 −0.237
Gold 0.36 −0.827
Construction 1.27 0.062
Air transport 1.80 0.348
Trucking 1.31 0.098
Consumer durables 1.44 0.132

27
 Portfolio Construction and the Single-Index Model 1

The single-index model creates a framework that separates these two quite different sources of return
variation and makes it easier to ensure consistency across analysts. We can lay down a hierarchy of the
preparation of the input list using the framework of the single-index model.
1. . Macroeconomic analysis is used to estimate the risk premium and risk of the market index.
2. Statistical analysis is used to estimate the beta coefficients of all securities and their residual variances,
σ2 (ei).
3. The portfolio manager uses the estimates for the market-index risk premium and the beta coefficient of a
security to establish the expected return of that security absent any contribution from security analysis.
The market-driven expected return is conditional on information common to all securities, not on
information gleaned from security analysis of particular firms. This market-driven expected return can
be used as a benchmark.
4. Security-specific expected return forecasts (specifically, security alphas) are derived from various
security-valuation models. Thus, the alpha value distills the incremental risk premium attributable to
private information developed from security analysis

28
Portfolio Construction and the Single-Index Model 2

Single-index model input list: The single-index model allows

Risk premium on the S&P 500 portfolio. us to solve for the optimal risky
Standard deviation of the S&P 500 portfolio. portfolio directly and to gain
insight into the nature of the
n sets of estimates of: solution. First we confirm that
n 1
• Beta coefficients: β P   wiβi we easily can set up the
i 1 optimization process to chart
n 1 the efficient frontier in this
• Stock residual variances: σ (eP )   wi2 σ 2 (ei )
2
framework along the lines of
i 1 the Markowitz model.
n 1
• Alpha values:
α P   wi α i
i 1

29
Portfolio Construction and the Single-Index Model

Optimal risky portfolio in the single-index model.

• Objective is to select portfolio weights to maximize the Sharpe ratio of the
portfolio. Optimal risky portfolio
n 1 n 1
E ( RP )  α P  E ( RM )β P   wi α i  E ( RM ) wiβi
in the single-index
model is a combination
i 1 i 1 1/2 of two component
 2  n 1

2 n 1  portfolios:
 P  [    σ (eP )]
2
P
2
M
2 1/2
 σ M   wi βi    wi σ (ei ) 
2 2
• Active portfolio,
  i 1  i1  denoted by A.
E ( RP ) • Market-index
SP  portfolio (i.e.,
P passive portfolio),
denoted by M

30
Summary of Optimization Procedure 1

1. Compute the initial position of each security:

αi
wi0 
σ 2 (ei )
2. Scale those initial positions:

wi0
wi  n

 i
w 0

i 1

3. Compute the alpha of the active portfolio:

n
α A   wi α i
i 1

31
Summary of Optimization Procedure 2

4. Compute the residual variance of A:

n
σ (eA )   wi2 σ 2 (ei )
2

i 1

5. Compute the initial position in A:

α A σ 2 (e A )
w 
0
A
E ( RM ) σ 2M
6. Compute the beta of A:

n
β A   wiβi
i 1

32
Summary of Optimization Procedure 3

7. Adjust the initial position in A:

wA0
w 
*
A
1  (1  β A )  wA0
Note: When β A  1  w*A  w0A
8. Optimal risky portfolio now has weights:

wM*  1  w*A
wi*  w*A  wi

33
Summary of Optimization Procedure 4

9. Calculate the risk premium of P (Optimal risky portfolio):

E ( RP )  ( wM*  w*Aβ A )  E ( RM )  w*Aα A

10. Compute the variance of P:

σ 2P  ( wM*  w*Aβ A ) 2 σ 2M  [ w*A σ(eA )]2

34
Optimal Risky Portfolio: Information Ratio

The optimal risky portfolio is the portfolio that maximizes the Sharpe ratio (risk-adjusted return) when
combined with a risk-free asset. In the context of active portfolio management, the Information Ratio (IR)
plays a crucial role in determining the best combination of active and passive investments.
Information Ratio.
• The contribution of the active portfolio depends on the ratio of its alpha to its residual standard deviation
• Calculated as the ratio of alpha to the standard deviation of diversifiable risk.
• The information ratio measures the extra return we can obtain from security analysis.

35
Optimal Risky Portfolio: Sharpe Ratio

The Sharpe ratio of an optimally constructed risky portfolio will exceed that of the index portfolio (the
passive strategy):

2
 αA 
S S 
2
P
2
M 
σ(
 A e )

36
Figure 8.4 Efficient Frontier: Index Model and Full-Covariance Matrix

Figure 8.4 Efficient frontier constructed from the index model and the full covariance matrix

Access the text alternative for slide images.

37
Table 8.4 Portfolios From the Index and Full-Covariance Models

Index Model Full-Covariance Model

A. Weights in Optimal Risky Portfolio
Market index 0.82 0.90
WMT (Walmart) 0.13 0.17
TGT (Target) −0.07 −0.14
VZ (Verizon) −0.05 −0.18
T (AT&T) 0.10 0.19
F (Ford) 0.07 0.08
GM (General Motors) 0.01 −0.03
B. Portfolio Characteristics
Risk premium 0.0605 0.0639
Standard deviation 0.1172 0.1238
Sharpe ratio 0.5165 0.5163

38
Is the Index Model Inferior to the Full-Covariance Model?

Factor Single Index Model (SIM) Full-Covariance Model

Assumes only market risk
Captures all relationships
Accuracy matters; ignores sector & firm-
between assets
specific relationships
Faster, requires estimating only N Slow, requires N(N−1)/2N(N-
Computational Efficiency
betas + 1 market variance 1)/2N(N−1)/2 covariances
May overlook diversification
Accurately identifies
Portfolio Diversification benefits from low-correlated
diversification opportunities
assets
Requires a large dataset of
Data Requirements Only needs market index data
historical correlations
Preferred for portfolios with Becomes computationally
Usability for Large Portfolios
hundreds/thousands of assets intensive as N increases

39
 Conclusion: Not Necessarily Inferior, Just
a Trade-Off
When the Index Model is Preferable:
✅ Large portfolios (e.g., mutual funds, ETFs) → Easier to implement.
✅ Market-driven stocks → Works well for stocks highly correlated with the market.
✅ Computational efficiency matters → Reduces complexity from covariance calcula ons.
When the Full-Covariance Model is Better:
✅ Small or sector-specific portfolios → Captures stock-to-stock correlations.
✅ Non-market risks matter → Suitable when industry/firm-specific risks are significant.
✅ Greater precision needed → Be er for hedge funds & ins tu onal investors.

•If accuracy is the priority, the Full-Covariance Model is superior.

•If simplicity & efficiency matter, the Index Model is a practical alternative, especially for large
portfolios.

40