Scribd
Upload
16 views

3.1 - Index Models

The document discusses the single-index model as an alternative to the Markowitz procedure for portfolio optimization. The single-index model requires far fewer estimates by relating security returns to a single market index. It estimates alpha, beta, and the market risk premium rather than individual security covariances. While requiring fewer estimates, it assumes residuals are uncorrelated and ignores industry effects.

Uploaded by

Muhammad Muhaimin
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
16 views

3.1 - Index Models

The document discusses the single-index model as an alternative to the Markowitz procedure for portfolio optimization. The single-index model requires far fewer estimates by relating security returns to a single market index. It estimates alpha, beta, and the market risk premium rather than individual security covariances. While requiring fewer estimates, it assumes residuals are uncorrelated and ignores industry effects.

Uploaded by

Muhammad Muhaimin
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 13

BMAN20072 Investment Analysis

Week 3-1: Index Models

BKM Ch. 8.1-8.5


Index Models
Overview

• How to design an optimal portfolio?


– Capital allocation
• How to optimally allocate capital between risk-free
and risky assets? [Week 1]
– Asset allocation
• How to optimally allocate budget across risky
assets?
• Through Markowitz procedure [Week 2]
• Alternative to the Markowitz procedure? [This
video]
1
Key terms

• Single index model


• Macroeconomic factor and security specific
factor
• Systematic risk premium
• Beta
• Alpha; Non-market premium (mispricing)

2
Drawback of the Markowitz
Procedure

• In our Portfolio Choice Assignment, we have


only 5 stocks.
• Input list of the Markowitz procedure:
– n = 5 estimates of expected returns
– n = 5 estimates of variances
– 𝑛2 − 𝑛 Τ2 = 10 estimates of covariances
– Total 20 estimates

• Ok, this is not bad…


3
Too many estimates needed

…But imagine you invest in 50 stocks:


• 50 estimates of expected excess returns
• 50 estimates of variances
• 𝑛2 − 𝑛 Τ2 = 1225 estimates of covariances
• Total 1325 estimates

4
Drawback of the Markowitz
Procedure

• …But imagine you invest in 50 stocks:


– 50 estimates of expected returns
– 50 estimates of variances
– 𝑛2 − 𝑛 Τ2 = 1225 estimates of covariances
– Total 1325 estimates

• If 100 stocks, it will be 5150 estimates!


• If 3000 stocks, 4.5 million!!
• Too many inputs!!!
5
The Single-Index Model

We need an alternative, feasible approach:


• The Single-Index Model
– With this approach, 3𝑛 + 2 estimates are
enough
• For 5 stocks: 17 rather than 20 estimates needed
for Markowitz procedure
• For 50 stocks: 152 rather than 1325!!
• For 3000 stocks: 9002 rather than 4.5 million!!!

• Too good to be true? What is the trick?


6
The Single-Index Model

• We estimate the regression equation for the


single-index model:
𝑅𝑖 𝑡 = 𝛼𝑖 + 𝛽𝑖 𝑅𝑀 𝑡 + 𝑒𝑖 (𝑡)
where
𝑅𝑖 = 𝑟𝑖 − 𝑟𝑓 : Excess return of a security i
𝑅𝑀 = 𝑟𝑀 − 𝑟𝑓 : Excess return of market index
ex) FTSE ALL Index
𝑒𝑖 : Firm-specific factor.
In regression, it is called the residual.
Assuming 𝐸 𝑒𝑖 = 0 𝑎𝑛𝑑 𝐶𝑜𝑣 𝑟𝑀 , 𝑒𝑖 = 0.
For 𝑖 ≠ 𝑗, 𝐶𝑜𝑣 𝑒𝑖 , 𝑒𝑗 = 0. 7
The Single-Index Model

• Expected excess return – beta relationship:


𝐸(𝑅𝑖 ) = 𝛼𝑖 + 𝛽𝑖 𝐸(𝑅𝑀 )

• 𝛽𝑖 : security’s sensitivity to the index, beta


• 𝛽𝑖 𝐸(𝑅𝑀 ): systematic risk premium
• 𝛼𝑖 : nonmarket premium (or mispricing), alpha
▪ 𝛼𝑖 > 0: security is under-priced
▪ 𝛼𝑖 < 0: security is over-priced
You believe there are some inefficiency in the market.

8
Risk and Covariance

• Total risk = Systematic risk + Firm-specific risk


𝜎𝑖2 = 𝛽𝑖2 𝜎𝑀2
+ 𝜎 2 (𝑒𝑖 )
• Covariance = Product of betas x Market-index
risk
2
𝐶𝑜𝑣 𝑟𝑖 , 𝑟𝑗 = 𝛽𝑖 𝛽𝑗 𝜎𝑀

• Correlation = Product of correlations with the market


index
2 2 2
𝛽𝑖 𝛽𝑗 𝜎𝑀 𝛽𝑖 𝜎𝑀 𝛽𝑗 𝜎𝑀
𝐶𝑜𝑟𝑟 𝑟𝑖 , 𝑟𝑗 = =
𝜎𝑖 𝜎𝑗 𝜎𝑖 𝜎𝑀 𝜎𝑗 𝜎𝑀
= 𝐶𝑜𝑟𝑟 𝑟𝑖 , 𝑟𝑀 × 𝐶𝑜𝑟𝑟(𝑟𝑗 , 𝑟𝑀 )
9
Expected Excess Return

We use index model to


𝐸(𝑅𝑖 ) = 𝛼𝑖 + 𝛽𝑖 𝐸(𝑅𝑀 )

• We need estimates for 𝛼𝑖 , 𝛽𝑖 , and 𝐸(𝑅𝑀 )


• 𝐸(𝑅𝑀 ): Macroeconomic analysis
• 𝛼𝑖 : Security Analysis
• 𝛽𝑖 : Regression with past data.
– Security’s sensitivity to macroeconomic factor is
relatively stable and persistent.

10
The Set of Estimates

3𝑛 + 2 estimates:
• n estimates of the extra-market expected
returns, 𝛼𝑖
• n estimates of the sensitivity coefficients, 𝛽𝑖
• n estimates of the firm-specific variances, 𝜎 2 (𝑒𝑖 )
• 1 estimate for the market risk premium, 𝐸(𝑅𝑀 )
• 1 estimate for the variance of the
2
macroeconomic factor, 𝜎𝑀

11
Pros and Cons

Pros:
• Less estimates needed
• Allows specialization of effort in security analysis.
Cons:
• Oversimplification of real-world uncertainty
• Ignores industry specific events:
– The model assumes correlation of residuals is always
zero.
– But residuals of stocks in same industry can be
correlated.
– Pair-wise covariance estimation takes this into
account. 12

You might also like