Lecture 4 Index Models 4.1 Markowitz Portfolio Selection Model
Lecture 4 Index Models 4.1 Markowitz Portfolio Selection Model
page 2 of 4
Markowitz Portfolio Selection Model
• Computationally intensive
– Even if
▷ Only a relatively small numbers of assets are involved
▷ The process is automated.
– Reason:
▷ Estimating n expected returns.
▷ Estimating n variances.
▷ Estimating (n2 − n)/2 covariances.
• Example.
– 10 stocks?
– 50 stocks?
– 100 stocks?
page 3 of 4
Markowitz Portfolio Selection Model
page 4 of 4
4.2 A Single-Factor Security Market
A Single-Factor Security Market
• Advantages
– Reduces the number of inputs for diversification.
– Easier for security analysts to specialize.
• Model
ri = E(ri) + ei
– ri: Rate of return on security i.
– E(ri): Expected component.
– ei: Unexpected component.
• We know from previous lecture
– The covariances between security returns tend to be positive.
– This stems from economic factors having common influences on
asset prices.
page 2 of 24
A Single-Factor Security Market
page 3 of 24
A Single-Factor Security Market
page 4 of 24
The Single-Index Model
• How we measure the unexpected macro factor m?
– Use return on a broad index of securities M as a proxy.
– ASX All Ordinaries Index (Australia).
– S&P 500 index (US).
• Such a model is called the Single-Index model.
• Formally
Rit = αi + βiRM t + eit
– αi: The security’s excess return when the market’s excess return is
zero.
– βi: The security’s sensitivity to the index.
– ei: The residual, or the unexpected firm-specific component of the
return.
page 5 of 24
The Single-Index Model
page 6 of 24
Concept Check 8.1
The data below describe a three-stock financial market that satisfies
the single-index model.
Stock Capitalization Beta Mean Excess Return Std. Dev.
A $3000 1.0 10% 40%
B $1940 0.2 2% 30%
C $1360 1.7 17% 50%
page 7 of 24
Concept Check 8.1
The data below describe a three-stock financial market that satisfies
the single-index model.
Stock Capitalization Beta Mean Excess Return Std. Dev.
A $3000 1.0 10% 40%
B $1940 0.2 2% 30%
C $1360 1.7 17% 50%
page 8 of 24
Concept Check 8.1
The data below describe a three-stock financial market that satisfies
the single-index model.
Stock Capitalization Beta Mean Excess Return Std. Dev.
A $3000 1.0 10% 40%
B $1940 0.2 2% 30%
C $1360 1.7 17% 50%
page 9 of 24
Concept Check 8.1
The data below describe a three-stock financial market that satisfies
the single-index model.
Stock Capitalization Beta Mean Excess Return Std. Dev.
A $3000 1.0 10% 40%
B $1940 0.2 2% 30%
C $1360 1.7 17% 50%
page 10 of 24
Concept Check 8.1
The data below describe a three-stock financial market that satisfies
the single-index model.
Stock Capitalization Beta Mean Excess Return Std. Dev.
A $3000 1.0 10% 40%
B $1940 0.2 2% 30%
C $1360 1.7 17% 50%
page 11 of 24
specific variance of B equals
σ 2(eB ) = σB
2
− βB σM = 0.302 − 0.22 × 0.252 = 0.0875
2 2
page 12 of 24
Concept Check 8.1
Suppose that the index model for the excess return of stocks A and B
is estimated with the following results
RA = 1.0% + 0.9RM + eA
RB = −2.0% + 1.1RM + eB
σM = 20%
σ(eA) = 30%
σ(eB ) = 10%
Find the standard deviation of each stock and the covariance between
them.
Solution:
The variance of each stock is β 2σM
2 + σ 2(e).
page 14 of 24
Advantages of The Single-Index Model
• What we need? Suppose there are n securities.
– n estimates of expected excess returns, E(Ri).
– n estimates of sensitivity coefficients βi.
– n estimates of the firm-specific variances σ 2(ei).
– 1 estimates for the common macroeconomic factor σM
2 .
• In total 3 × n + 1.
• Remember in the Markowitz Portfolio Selection we need 0.5 × n2 +
1.5 × n.
• The single-index model provides further insight by recognizing that
different firms have different sensitivities to macroeconomic events.
• The model also summarizes the distinction between macroeconomic
and firm-specific risk factors.
page 15 of 24
The Index Model and Diversification
• Portfolio
Rp = αp + βpRM + ep
• The number of stocks included in the portfolio increases, the part of the
portfolio risk attributable to non-market factors becomes ever smaller.
• Market risk remains, regardless of the number of firms combined into
the portfolio.
Xn X
n
Rp = ωi R i = ωi(αi + βiRM + ei)
i=1 i=1
Xn Xn X
n
= ωi α i + ( ωiβi)RM + ωiei
i=1 i i=1
Xn
1 2
2
σ (ep) = ωi2σ 2(ei) |{z}
= σ̄ (e)
When ω = 1/n
n
i=1 i
page 16 of 24
The Index Model and Diversification
page 17 of 24
Estimating the Single-Index Model
• Using historical data to estimate the single-index model yields the Se-
curity Characteristic Line (SCL).
• Formally, we regress the monthly excess return for firm i against the
monthly excess return on a broad index.
Ri,t = α + βRM,t + ei,t
– α: The intercept is the asset’s alpha for the sample period.
– β : The slope represents the beta of the asset.
– ei,t: The error terms represent the difference between actual returns
and those predicted by the regression line, or the unexpected firm-
specific component of the return.
page 18 of 24
Estimating the Single-Index Model
page 19 of 24
Estimating the Single-Index Model
page 20 of 24
Estimating the Single-Index Model
page 21 of 24
Estimating the Single-Index Model
• Preparing data
Return − Risk-free rate → Excess Return
• Regression
Rit = α + βRM t + ϵit
– Ordinary Least Squares
PT
i=1(RM t − R̄M t)(Rit − R̄it)
β= PT
i=1(RM t − R̄M t)
2
α = R̄it − β R̄M t
• Excel example
page 22 of 24
The Single-Index Model
page 23 of 24
The Single-Index Model
page 24 of 24
4.3 Portfolio Construction and the Single-Index Model
Portfolio Construction and the Single-Index Model
• We use alpha, beta, and risk and return estimates for the market index
to generate efficient frontier.
– Alpha, beta and index risk premium estimates can be used to gen-
erate expected returns.
E(Ri) = αi + βiE(RM )
– Beta, residual variance and index variance estimates can be used to
construct the covariance matrix by recalling.
σi2 = βi2σM2
+ σ 2(ei)
2
Cov(ri, rj ) = βiβj σM
– We can then identify the optimal risky portfolio by maximizing the
Sharpe ratio whilst ensuring weights held in all risky assets sum to
1.
page 2 of 9
Portfolio Construction and the Single-Index Model
• Specifically
∑
n ∑
n
E(Rp) = αp + E(RM )βp = ωiαi + E(RM ) ωiβi
i i=1
( n )2 1
2
∑ ∑
n
+ σ (ep)] = σM ωi2σ 2(ei)
1
σp = [βp2σM
2 2 2 2
ωiβi +
i=1 i=1
Sp =
E(Rp)
σp
page 3 of 9
The Optimal Risky Portfolio and the Single-Index Model
• When returns follow the index model, we can directly solve for the
optimal risky portfolio.
• To do this, we must view the portfolio as a combination of
– An active portfolio, A, comprising the n analyzed securities with
significant alphas in some weight.
– The market-index portfolio, M, which is a passive portfolio.
page 4 of 9
The Optimal Risky Portfolio and the Single-Index Model
• The weight in which the active portfolio will be held in the optimal
risky portfolio will
– Reflect both its contribution to the optimal risky portfolio (i.e. via
its alpha).
– Also take into account how it contributes to the risk of the optimal
risky portfolio (i.e. via its residual variance).
• The Sharpe ratio of an optimally constructed risky portfolio will be
greater than that enjoyed by the index (passive portfolio).
[ ]2
2 2 αA
Sp = SM +
σ(eA)
– The extra return provided by security analysis will be captured by
the information ratio.
page 5 of 9
Summary of Optimization Procedure
1. Compute the initial position of each security as ωi0 = αi/σ 2(ei).
2. Scale those initial positions so that
ωi0
ωi = ∑n 0
ω
i=1 i
∑n
3. Compute the alpha of the portfolio as αA = i=1 ωiαi.
∑n
4. Compute the residual variance as σ 2(eA) = ω
i=1 i
2σ 2(e ).
i
page 6 of 9
Summary of Optimization Procedure
∗ = 1 − ω ∗ and ω ∗ =
7. The optimal risky portfolio now has weights ωM A i
∗ω .
ωA i
page 7 of 9
The Index Model in Practice
page 8 of 9
The Index Model in Practice
• It is also important to recognise that
– Betas tend towards one over time.
– This trend suggests betas estimated using historical data may not
provide the best estimate of future betas.
– Therefore, it may be necessary to forecast betas in some manner.
• Empirical studies offer some guidance on useful predictors of betas.
– Variance of earnings and cash flows.
– Growth in EPS.
– Firm size.
– Dividend yield.
– Firm leverage, as measured by the ratio of debt to assets.
– Industry in which the firm operates.
page 9 of 9