Chapter 6A: Diversification

Essential reading

• Bodie, Z., A. Kane and A.J. Marcus Investments. (Boston, Mass.; London: McGraw-Hill Irwin,
2008) Chapter 6 "Capital Allocation to Risky Assets", Chapter 7 "Optimal Risky Portfolios",
Chapter 8 "Index Models", Chapter 9 "The Capital Asset Pricing Model", Chapter 10
"Arbitrage Pricing Theory and Multifactor Models of Risk and Return" and Chapter 27 "The
Theory of Active Portfolio Management".
• Fabozzi, F. J. and H. M. Markowitz (eds) The Theory and Practice of Investment
Management. (Hoboken, NJ: John Wiley & Sons, 2011) Chapters 3, 4 and 5.

Further reading
• Campbell, J.Y. and L.M. Viceira Strategic Asset Allocation. (New York: Oxford University
Press) Chapter 2.

2
Learning Outcomes

• Compute the expected return, risk premium, variance and the standard deviation of a risky portfolio
• Compute the expected return, risk premium, variance and the standard deviation of a combination of a risky
portfolio and a risk free asset
• Define mean-variance preferences and explain how investors with mean-variance preferences choose
portfolios
• Derive the optimal portfolio with the knowledge of the investors' risk aversion coefficient
• Define the portfolio frontier
• Explain how the existence of a portfolio frontier on which investors choose their optimal portfolios implies that
the CAPM pricing formula holds
• Define the concepts of systematic and unsystematic risk, and explain how these concepts are used to
simplify the problem of estimating the covariance between asset returns
• Discuss the implications of the single index model to effective diversification in detail
• Review the Treynor-Black model
• Define factor models, and illustrate well-established pricing formulas
• Explain why myopic portfolio choice may sometimes be optimal even though the investors have a long
investment horizon.

3
Chapter 6A

• Expected portfolio return and variance

• Definition of risk premium
• Numerical example
• Asset allocation: Two assets
• Mean-variance preferences
• CARA utility and normal returns

4
Chapter 6B and 6C

• Optimal asset allocation with a risk free asset

• CAPM – Capital Assets pricing model
• The portfolio frontier
• Expected returns relationships
• Estimation issues
• Diversification: The single index model
• The Treynor-Black model
• Factor models
• Myopic portfolio choice and asset allocation over investment horizon

5
Expected portfolio return and variance

• Expected Portfolio return

• Portfolio Variance
Expected portfolio return

1. Determine the portfolio weight(𝑤𝑖 ) (percentage

allocation) of each assets.
𝑉𝑎𝑙𝑢𝑒 𝑜𝑓 𝑎𝑠𝑠𝑒𝑡 𝑖
𝑤𝑖 =
𝑇𝑜𝑡𝑎𝑙 𝑃𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 𝑉𝑎𝑙𝑢𝑒

• The weights should add up to 100%

σ𝑖 𝑤𝑖 = 1
Example
• Invest $30,000 in Tesla, $50,000 in MSFN and
$20,000 in Meta, your portfolio consists of
20,000
• 𝑤𝑇𝑆𝐿𝐴 = = 0.20 𝑜𝑟 20%
20000+50000+30000
50,000
• 𝑤𝑀𝑆𝐹𝑁 = = 0.50 𝑜𝑟 50%
20000+50000+30000
30,000
• 𝑤𝑀𝐸𝑇𝐴 = = 0.30 𝑜𝑟 30%
20000+50000+30000

7
Expected portfolio return

2. Find the expected return of each individual asset(𝑟𝑖 ).

• based on historical returns, forecasted economic conditions, and risk profile of the asset, etc.

𝑝1 + 𝐷𝑖𝑣1 − 𝑝0
𝐻𝑜𝑙𝑑𝑖𝑛𝑔 𝑃𝑒𝑟𝑖𝑜𝑑 𝑅𝑒𝑡𝑢𝑟𝑛 =
𝑝0
249−170
• 𝐻𝑃𝑅𝑇𝑆𝐿𝐴 = 170
= 46.47%

335−275
• 𝐻𝑃𝑅𝑀𝑆𝐹𝑁 = = 21.82%
275

298−181
• 𝐻𝑃𝑅𝑀𝐸𝑇𝐴 = 181
= 64.64%

8
Expected portfolio return

9
Expected portfolio return

Date TSLA MSFT META

Jan-21 12.45% 4.29% -5.43%
Feb-21 -14.87% 0.41% -0.27%
Mar-21 -1.12% 1.46% 14.33%
Apr-21 6.21% 6.96% 10.37%
May-21 -11.87% -0.76% 1.12%
Jun-21 8.71% 8.50% 5.77%
Jul-21 1.10% 5.17% 2.47%
Aug-21 7.06% 6.16% 6.48%
Sep-21 5.40% -6.61% -10.54%
Oct-21 43.65% 17.63% -4.66%
Nov-21 2.76% -0.13% 0.28%
Dec-21 -7.69% 1.73% 3.66%
10
Expected portfolio return

Jan-22 -11.36% -7.53% -6.86%

Feb-22 -7.08% -3.72% -32.63%
Mar-22 23.80% 3.19% 5.37%
Apr-22 -19.19% -9.99% -9.84%
May-22 -12.92% -1.81% -3.41%
Jun-22 -11.19% -5.53% -16.73%
Jul-22 32.38% 9.31% -1.33%
Aug-22 -7.25% -6.67% 2.41%
Sep-22 -3.76% -10.93% -16.72%
Oct-22 -14.22% -0.33% -31.34%
Nov-22 -14.43% 10.22% 26.77%
Dec-22 -36.73% -6.00% 1.90%

11
Expected portfolio return

Jan-23 40.62% 3.33% 23.79%

Feb-23 18.76% 0.90% 17.43%
Mar-23 0.85% 15.59% 21.15%
Apr-23 -20.80% 6.58% 13.39%
May-23 24.11% 7.11% 10.15%
Jun-23 28.36% 3.70% 8.41%
Jul-23 2.16% -1.36% 11.02%
Aug-23 -3.50% -2.22% -7.13%
Sep-23 -3.05% -3.66% 1.46%
Oct-23 -19.73% 7.08% 0.35%
Nov-23 10.63% 6.63% 5.83%

12
Expected portfolio return

TSLA MSFT META

Arithmetic average return(monthly) 1.38% 1.68% 1.34%
Arithmetic average return(Annual) 16.55% 20.12% 16.12%

• 𝑟𝑇𝑆𝐿𝐴 = 16.55%
• 𝑟𝑀𝑆𝐹𝑁 = 20.12%
• 𝑟𝑀𝐸𝑇𝐴 = 16.12%

13
Expected portfolio return

3. Multiply the expected return of each asset by its weight in the portfolio.
• Important formula for this chapter

𝐸 𝑟𝑝 = σ𝑖 𝑤𝑖 𝐸(𝑟𝑖 )

where

• 𝐸 𝑟𝑝 : expected return of the portfolio:

• 𝑤𝑖 : weight of asset 𝑖 in the portfolio

• 𝐸(𝑟𝑖 ): expected return on asset 𝑖
In our example, the expected return is given by:

• 𝐸 𝑟𝑝 = 20% ∙ 16.55% + 50% ∙ 20.12% + 30% ∙ 16.12% = 18.21%

𝑟𝑇𝑆𝐿𝐴 = 16.55%
𝑟𝑀𝑆𝐹𝑁 = 20.12%
𝑟𝑀𝐸𝑇𝐴 = 16.12%
14
Portfolio variance

• Variance of the return on a portfolio

• where we sum over all assets 𝑖 = 1,2, … , 𝑛 and 𝑗 = 1,2, … , 𝑛

• The variance of a portfolio of assets is the sum of the variances of the assets multiplied by
the portfolio weights squared, plus two times the sum of all covariances multiplied by the
two corresponding portfolio weights.
Example: Two assets 1, 2

𝑉𝑎𝑟 𝑟𝑝 = 𝑤12 𝑉𝑎𝑟 𝑟1 + 𝑤22 𝑉𝑎𝑟 𝑟2 + 𝑤1 𝑤2 𝐶𝑜𝑣 𝑟1 , 𝑟2 + 𝑤2 𝑤1 𝐶𝑜𝑣(𝑟2 , 𝑟1 )

• Since 𝐶𝑜𝑣 𝑟1, 𝑟2 = 𝐶𝑜𝑣(𝑟2, 𝑟1 ),

𝑉𝑎𝑟 𝑟𝑝 = 𝑤12 𝑉𝑎𝑟 𝑟1 + 𝑤22 𝑉𝑎𝑟 𝑟2 + 2𝑤1 𝑤2 𝐶𝑜𝑣 𝑟1 , 𝑟2

15
Portfolio variance

Example: Three assets

• 𝑉𝑎𝑟 𝑟𝑝 = 𝜎𝑝2 = 𝑤12 𝑉𝑎𝑟 𝑟1 + 𝑤22 𝑉𝑎𝑟 𝑟2 + 𝑤32 𝑉𝑎𝑟 𝑟3

+2𝑤1 𝑤2 𝐶𝑜𝑣 𝑟1 , 𝑟2
+2𝑤1 𝑤3 𝐶𝑜𝑣 𝑟1 , 𝑟3
+2𝑤2 𝑤3 𝐶𝑜𝑣 𝑟2 , 𝑟3
• We have 3 variance terms and 6 covariance terms

16
Portfolio variance

• In our example

j 0.2 0.5 0.3

i TSLA MSFT META
0.2 TSLA 0.397045 0.070131 0.074802
0.5 MSFT 0.070131 0.054635 0.053229
0.3 META 0.074802 0.053229 0.208446
• 𝑉𝑎𝑟 𝑟𝑝 = 0.22 (0.397045) + 0.52 (0.054635) + 0.32 (0.208446)
+2(0.2)(0.5)(0.070131)
+2(0.2)(0.3)(0.074802)
+2(0.5)(0.3) (0.053229)
= 0.087272

17
Portfolio standard deviation

Standard deviation is a measure of the dispersion or variation of a data

set from its mean.
In investment, the standard deviation measures the volatility of returns

𝑆𝐷 𝑟𝑝 = 𝑉𝑎𝑟(𝑟𝑝 ) = 𝜎𝑝 = 0.087272

= 0.295418 ≈ 29.54%

18
Exercise 1

• Calculate the portfolio return and portfolio variance given the following matrix

wi BT Lloyds Bank M&S Expected Return

BT 0.3 9.00% 5.00% 2.00% 10%
Lloyds Bank 0.5 5.00% 16.00% 3.00% 13%
M&S 0.2 2.00% 3.00% 8.00% 8%

19
Exercise 1

20
Correlation coefficient – Standardized Covariance

Note that the portfolio has a lower variance(except MSFT) than

any of the stocks individually.
Why?
The correlation between the stocks is imperfect, so that some
of the risk in one stock will be offset by the risk in the other stocks.
• Correlation coefficient is defined as
𝐶𝑜𝑣 𝑟1 ,𝑟2
𝜌12 =
𝜎1 𝜎2

• where 𝜌12 is the correlation coefficient between 𝑟1 𝑎𝑛𝑑 𝑟2 , and

𝜎1 and 𝜎2 are the standard deviations of 𝑟1 𝑎𝑛𝑑 𝑟2
21
Correlation coefficient – Standardized Covariance

• 𝜌12 = +1 indicates a perfect positive relationship.

• As one variable increases, the other variable also increases proportionally.

• 𝜌12 = −1 indicates a perfect negative relationship

• As one variable increases, the other variable decreases proportionally.

• 𝜌12 = 0 indicates no linear relationship between the variables.

• 0 < 𝜌12 < +1 : positive correlation coefficient means as one variable increases, the other
tends to increase.
• −1 < 𝜌12 < 0 A negative coefficient means as one increases, the other tends to decrease.
• Closer to either -1 or 1, the stronger the linear relationship between the variables.
• Closer to 0 indicates a weaker linear relationship.

22
Correlation coefficient

Example: Telsa and Microsoft

𝐶𝑜𝑣 𝑟1 ,𝑟2 0.070131
• 𝜌12 = = = 0.4762
𝜎1 𝜎2 0.397045 0.054635
• which implies that if you mix Telsa and Microsoft in a portfolio, some of the risk in the
individual stocks will be offset by the fact that the correlation coefficient is not perfect (i.e. it
is not 1).

Correlation
TSLA 1.0000 0.4762 0.2600
MSFT 0.4762 1.0000 0.5672
META 0.2600 0.4988 1.0000
23
Exercise 1

• Calculate the Correlation coefficient given the following matrix

wi BT Lloyds Bank M&S Expected Return

BT 0.3 9.00% 5.00% 2.00% 10%
Lloyds Bank 0.5 5.00% 16.00% 3.00% 13%
M&S 0.2 2.00% 3.00% 8.00% 8%

Correlaton
BT 1.0000 0.4167 0.2357
Lloyyds 0.4167 1.0000 0.2652
M&S 0.2357 0.2652 1.0000

24
Two-Assets Portfolio

• A portfolio that consist of a fraction 𝑤 invested in 𝐴 and the

remaining fraction 1 − 𝑤 invested in 𝐵, has expected return and
variance as follow:
𝐸(𝑟) = 𝑤𝐸(𝑟𝐴 ) + (1 − 𝑤)𝐸(𝑟𝐵 )
𝑉𝑎𝑟(𝑟) = 𝑤 2 𝑉𝑎𝑟(𝑟𝐴 ) + 2𝑤(1 − 𝑤)𝐶𝑜𝑣(𝑟𝐴 , 𝑟𝐵 ) + 1 − 𝑤 2 𝑉𝑎𝑟(𝑟𝐵 )
𝑉𝑎𝑟 𝑟 = 𝑤 2 𝜎𝐴2 + 2𝑤(1 − 𝑤)𝜌𝐴𝐵 𝜎𝐴 𝜎𝐵 + 1 − 𝑤 2 𝜎𝐵2

25
Two-Assets Portfolio

• If 𝐸 𝑟𝐴 = 𝐸(𝑟𝐵 ), an investor with mean-variance preferences chooses to hold A and B such

that the variance is minimized.
• The minimum variance portfolio can be found as follow:

min 𝑉𝑎𝑟 𝑟 = 𝑤 2 𝜎𝐴2 + 2𝑤(1 − 𝑤)𝜌𝐴𝐵 𝜎𝐴 𝜎𝐵 + 1 − 𝑤 2 𝜎𝐵2

𝑤
𝑑𝑉𝑎𝑟(𝑟)
= 2𝑤𝜎𝐴2 + 2 − 4𝑤 𝜌𝐴𝐵 𝜎𝐴 𝜎𝐵 − 2 1 − 𝑤 𝜎𝐵2 = 0
𝑑𝑤

𝑤𝜎𝐴2 + 1 − 2𝑤 𝜌𝐴𝐵 𝜎𝐴 𝜎𝐵 − 1 − 𝑤 𝜎𝐵2 = 0

𝑤 𝜎𝐴2 + 𝜎𝐵2 − 2𝜌𝐴𝐵 𝜎𝐴 𝜎𝐵 = 𝜎𝐵2 − 𝜌𝐴𝐵 𝜎𝐴 𝜎𝐵
2
𝜎𝐵 −𝜌𝐴𝐵 𝜎𝐴 𝜎𝐵
𝑤= 2 +𝜎 2 −2𝜌
𝜎𝐴 𝐵 𝐴𝐵 𝜎𝐴 𝜎𝐵

𝜎𝐵 (𝜎𝐵 −𝜌𝐴𝐵 𝜎𝐴 )
𝑤=
𝜎𝐴 −𝜎𝐵 2 +2𝜎𝐴 𝜎𝐵 (1−𝜌𝐴𝐵 )
26
Example

• If we consider two assets A and B with expected return 15% and variance 1% and 9%,
respectively, with correlation coefficient -1,
• What is the weight of assets A and B to minimum portfolio variance?
• What is the minimum variance of the portfolio?
• min 𝑉𝑎𝑟 𝑟 = 𝑤 2 𝜎𝐴2 + 2𝑤(1 − 𝑤)𝜌𝐴𝐵 𝜎𝐴 𝜎𝐵 + 1 − 𝑤 2 𝜎𝐵2
𝑤

• 𝜌𝐴𝐵 = −1; 𝜎𝐴2 = 0.01; 𝜎𝐵2 = 0.09 ⇒ 𝜎𝐴 = 0.10; 𝜎𝐵 = 0.30

• min 𝑉𝑎𝑟 𝑟 = 0.01𝑤 2 + 2𝑤 1 − 𝑤 −1 (0.10)(0.30) + 1 − 𝑤 2 (0.09)
𝑤
𝑑
• 𝑉𝑎𝑟 𝑟 = 0.02𝑤 + 2 − 4𝑤 −0.03 + 0.09 −2 1 − 𝑤 = 0
𝑑𝑤

• 0.02𝑤 + 0.12𝑤 + 0.18𝑤 − 0.06 − 0.18 = 0

0.24
• 𝑤= = 0.75
0.32

27
 w 1-w Var_P Std_P rP
Example -0.40
-0.35
1.40
1.35
21.16%
19.36%
46.00%
44.00%
15.00%
15.00%
-0.30 1.30 17.64% 42.00% 15.00%
-0.25 1.25 16.00% 40.00% 15.00%
• The solution can also be found by using -0.20 1.20 14.44% 38.00% 15.00%
-0.15 1.15 12.96% 36.00% 15.00%
the equation -0.10 1.10 11.56% 34.00% 15.00%
-0.05 1.05 10.24% 32.00% 15.00%
0.00 1.00 9.00% 30.00% 15.00%
0.05 0.95 7.84% 28.00% 15.00%
0.10 0.90 6.76% 26.00% 15.00%
0.15 0.85 5.76% 24.00% 15.00%
0.30(0.30−(−1)(0.10) 0.12 0.20 0.80 4.84% 22.00% 15.00%
• 𝑤= = = 0.25 0.75 4.00% 20.00% 15.00%
0.1−0.3 2 +2(0.10)(0.30)(1− −1 ) 0.16
0.30 0.70 3.24% 18.00% 15.00%
0.75 0.35 0.65 2.56% 16.00% 15.00%
0.40 0.60 1.96% 14.00% 15.00%
• Check the variance of the portfolio 0.45 0.55 1.44% 12.00% 15.00%
0.50 0.50 1.00% 10.00% 15.00%
• 𝑉𝑎𝑟 𝑟 = 0.752 0.01 + 0.55 0.45 0.64% 8.00% 15.00%
2 0.75 0.25 −1 0.10 0.30 + 0.60 0.40 0.36% 6.00% 15.00%
0.252 0.09 = 0 0.65
0.70
0.35
0.30
0.16%
0.04%
4.00%
2.00%
15.00%
15.00%
0.75 0.25 0.00% 0.00% 15.00%
0.80 0.20 0.04% 2.00% 15.00%
0.85 0.15 0.16% 4.00% 15.00%
0.90 0.10 0.36% 6.00% 15.00%
0.95 0.05 0.64% 8.00% 15.00%
1.00 0.00 1.00% 10.00% 15.00%
28
Example

29
Power of diversification in portfolio management.

• Even when two assets have the same expected rate of return but
different variances, it's possible to reduce portfolio variance
significantly by combining them if their correlation coefficient is
less than 1.
• When the correlation coefficient is -1, we can find a specific
weights that create a portfolio with zero variance. This portfolio is
often referred to as a "perfectly hedged portfolio.

30
Two-Assets Portfolio

• If we consider two assets A and B with expected return 10% and variance 16% and 9%,
respectively, with correlation coefficient 0.5, the minimum variance portfolio consists of:

31
Exercise 1

• Chris is considering forming a portfolio using stocks A and B. The expected returns and
standard deviations are listed below

Expected Return Standard Deviation

Stock A 5% 10%

Stock B 10% 18.5%

• The return correlation between the two stocks is 0.105.

• Should Chris invest in a portfolio consisting of 60% stock A and 40% stock B, or invest
everything in stock A?
• Does your choice depend on your assumption about risk aversion?

32
Exercise 1

• Expected return on the portfolio is:

𝐸 𝑟 = 𝑤𝐸 𝑟𝐴 + 1 − 𝑤 𝐸 𝑟𝐵 = 0.6 ∙ 0.05 + 0.4 ∗ 0.1 = 0.07 𝑜𝑟 7%
• Variance of the portfolio returns is:

𝑉𝑎𝑟 𝑟 = 𝑤 2 𝜎𝐴2 + 2𝑤 1 − 𝑤 𝜌𝐴𝐵 𝜎𝐴 𝜎𝐵 + 1 − 𝑤 2 𝜎𝐵2

2 2 2 2
• 𝑉𝑎𝑟 𝑟 = 0.6 0.1 + 2 0.6 0.4 0.1 0.185 0.105 + 0.4 0.185 = 0.01

• 𝑆𝐷 𝑟 = 0.01 = 0.1 𝑜𝑟 10%

• The 60/40 split portfolio has the same risk as stock A but a higher expected return.
• Therefore, Chris should choose the portfolio regardless of risk appetite/aversion.

33
Exercise 1

34
Definition of risk premium
Definition of risk premium

• Investors demand higher returns to take on more risk.

This phenomenon where investors dislike risk is called
risk aversion.
• For investors to be willing to hold risky assets, those
assets must offer returns above the "risk-free" return.
𝑅𝑖𝑠𝑘 𝑝𝑟𝑒𝑚𝑖𝑢𝑚 = 𝐸(𝑟) − 𝑟𝐹
• 𝐸(𝑟): the expected rate of return on the asset
• 𝑟𝐹 : the risk free return.

36
Definition of risk premium

• In specific cases we know what this risk premium looks like.

• For instance, if the CAPM is true, the risk premium is:
Risk Premium under CAPM = 𝛽(𝐸 𝑟𝑀 − 𝑟𝐹 )
• where 𝛽 is the beta factor of the asset, and 𝐸(𝑟𝑀 − 𝑟𝐹 ) is the risk
premium on the market portfolio (or the market index).
• We will get back to the derivation of the CAPM model later on in
this chapter.

37
Asset allocation: Two assets
Asset allocation: Two assets

One risk-free asset and one risky asset

• Investor chooses a portfolio with weight 𝑤 in the risky asset and 1– 𝑤 in the
risk-free asset
• Return of the portfolio is 𝑟𝑝 = 𝑤𝑟 + (1 − 𝑤)𝑟𝐹
• Expected return on the portfolio is then given by:
𝐸 𝑟𝑝 = 𝑤𝐸 𝑟 + 𝑟𝐹 − 𝑤𝑟𝐹
= 𝑟𝐹 + 𝑤(𝐸 𝑟 − 𝑟𝐹 )
𝑟𝑖𝑠𝑘 𝑝𝑟𝑒𝑚𝑖𝑢𝑚
• i.e. the risk-free rate of return plus the risk premium of the risky asset times
the portfolio weight in the risky asset.

39
Asset allocation: Two assets

• 𝑟~(𝐸 𝑟 , 𝜎 2 ) and 𝑟𝐹 is a given constant, therefore, the variance of risk-free asset’s return is
zero. It also implies the 𝐶𝑜𝑣 𝑟, 𝑟𝐹 = 0
• The variance of the portfolio equals:
𝜎𝑝2 = 𝑉𝑎𝑟 𝑟𝑝 = 𝑤 2 𝜎 2 + 2𝑤 1 − 𝑤 𝜌𝜎 𝜎
ด𝐹 + 1 − 𝑤 2𝜎 2
ด𝐹 = 𝑤2𝜎 2
=0 =0

• If we work with standard deviation instead of variance, the expected return and the
standard deviation of return on the portfolio are both linear in the portfolio weight in the
risky asset:

𝐸 𝑟𝑝 = 𝑟𝐹 + 𝑤 𝐸 𝑟 − 𝑟𝐹 ; 𝜎𝑝 = 𝑤𝜎
𝜎𝑝
• This implies 𝑤 = , the 𝐸 𝑟𝑝 can be rewritten as
𝜎
𝐸 𝑟 −𝑟𝐹
𝐸 𝑟𝑝 = 𝑟𝐹 + 𝜎𝑝 the investment opportunity set
𝜎
𝑆𝑙𝑜𝑝𝑒

40
Asset allocation: Two assets

• The portfolios have a linear expected return and standard deviation in the weight on the
risky asset.
𝐸(𝑟𝑝 )

𝐸 𝑟
Δ𝑦 𝐸 𝑟 − 𝑟𝐹
𝑆𝑙𝑜𝑝𝑒 = =
Δ𝑥 𝜎
𝐸 𝑟 − 𝑟𝐹

41
Mean-variance preferences

• If investors have mean-variance preferences they have a utility function over portfolios that
take the form
𝑢(𝜇, 𝜎 2 )
• This function is increasing in 𝜇 and decreasing in 𝜎 2
• A rational investor picks a portfolio that maximises his utility, that is, a portfolio that has an
optimal risk–return trade– off.

42
Mean-variance preferences

Consider a general two-asset example

Portfolio expected return = 𝜇 = 𝑤𝜇1 + (1 − 𝑤)𝜇2
• and
• Portfolio variance 𝜎 2 = 𝑤 2 𝜎12 + 2𝑤(1 − 𝑤)𝜎12 + 1 − 𝑤 2 𝜎22
• The optimization problem is given by
max 𝑢(𝜇(𝑤), 𝜎 2 (𝑤))
𝑤

43
Mean-variance preferences

• The first order condition* is:

𝜕𝑢 𝜕𝜇 𝜕𝑢 𝜕𝜎2
∙ + ∙ =0
𝜕𝜇 𝜕𝑤 𝜕𝜎2 𝜕𝑤

𝜕𝑢 𝜕𝜇 𝜕𝑢 𝜕𝜎2
∙ = − 2 ∙
𝜕𝜇 𝜕𝑤 𝜕𝜎 𝜕𝑤
𝜕𝑢 𝜕𝜇 𝜕𝜇
𝜕𝜎2 𝜕𝑤
𝑀𝑈𝜎2 𝜕𝑤
𝜕𝑢 =− 𝜕𝜎2
⇔ =− 𝜕𝜎2
𝜕𝜇
ถ
𝑀𝑈𝜇
𝜕𝑤 𝜕𝑤
𝑀𝑅𝑆

MRS is the marginal rate of substitution between risk and return.

44
Mean-variance preferences
𝜇 = 𝑤𝜇1 + (1 − 𝑤)𝜇2
𝜕𝜇
= 𝝁𝟏 − 𝝁𝟐
𝜕𝑤
𝜎 2 = 𝑤 2 𝜎12 + 2𝑤(1 − 𝑤)𝜎12 + 1 − 𝑤 2 𝜎22
𝜕𝜎2
= 2𝑤𝜎12 + 2 − 4𝑤 𝜎12 + 2 1 − 𝑤 𝜎22
𝜕𝑤

2𝑤𝜎12 − 2𝑤𝜎22 + 2𝜎12 − 4𝑤𝜎12 + 2𝜎22

• Note that 𝜎12 = 𝜌12 𝜎1 𝜎2 and grouping 𝑤 gives
𝜕𝜎2
= 2𝑤 𝜎12 − 2𝜎12 − 𝜎22 + 2 𝜎12 − 𝜎22 = 𝟐𝒘 𝝈𝟐𝟏 − 𝟐𝝆𝟏𝟐 𝝈𝟏 𝝈𝟐 + 𝝈𝟐𝟐 + 𝟐 𝝆𝟏𝟐 𝝈𝟏 𝝈𝟐 − 𝝈𝟐𝟐
𝜕𝑤

𝜕𝜇 𝝁𝟐 − 𝝁𝟏
𝝁 − 𝝁
− 𝜕𝑤2 = −
𝟏 𝟐
𝜕𝜎 𝟐𝒘 𝝈𝟐𝟏 − 𝟐𝝆𝟏𝟐 𝝈𝟏 𝝈𝟐 + 𝝈𝟐𝟐 − 𝟐𝝈𝟐 𝝈𝟐 − 𝝆𝟏𝟐 𝝈𝟏
𝜕𝑤

45
Mean-variance preferences
𝜕𝑢 𝜕𝜇
𝜕𝜎 2 𝝁𝟐 − 𝝁𝟏 𝜕𝑤
𝑀𝑅𝑆 = = = −
𝜕𝑢 𝟐𝒘 𝝈𝟐𝟏 − 𝟐𝝆𝟏𝟐 𝝈𝟏 𝝈𝟐 + 𝝈𝟐𝟐 − 𝟐𝝈𝟐 𝝈𝟐 − 𝝆𝟏𝟐 𝝈𝟏 𝜕𝜎 2
𝜕𝜇 𝜕𝑤

The important point about this is that the investors choose

portfolios in order to balance expected return against
portfolio variance.
Everything else being equal, they choose portfolios that
have lower variance, or they choose portfolios that have
higher expected return.
46
Optimal asset allocation with a risk free asset

• CAPM tells us that investors pick portfolios on the portfolio frontier, and that the portfolio
frontier can be spanned by two known portfolios, the risk free asset and the market
portfolio.(two–fund separation).
• Therefore, for any arbitrary portfolio 𝑟𝑃 , there exists a mixture portfolio 𝑥𝑟𝑀 + (1– 𝑥)𝑟𝐹 such
that:
𝑟𝐹 + 𝑥 𝐸 𝑟𝑀 − 𝑟𝐹 ≥ 𝐸 𝑟𝑃 ;
𝑉𝑎𝑟(𝑟𝐹 + 𝑥 𝐸 𝑟𝑀 − 𝑟𝐹 ≤ 𝑉𝑎𝑟(𝑟𝑃 )
• where at least one inequality is strict unless 𝑟𝑃 is also a mixture portfolio of the risk free
asset and the market portfolio.

47
Optimal asset allocation with a risk free asset

• If we consider the optimal portfolios, therefore, we can set the parameters

• 𝜇1 = 𝐸(𝑟𝑀 ) and 𝜎1 = 𝜎𝑀 ;
• 𝜇2 = 𝑟𝐹 and 𝜎2 = 0

• We find that under the optimal allocation:

𝜕𝜇
𝜕𝑤 𝝁𝟐 −𝝁𝟏 𝑟𝐹 −𝐸 𝑟𝑀
− 𝜕𝜎2
= = 2
𝟐𝒘 𝝈𝟐𝟏 −𝟐𝝆𝟏𝟐 𝝈𝟏 𝝈𝟐 +𝝈𝟐𝟐 +𝟐 𝝆𝟏𝟐 𝝈𝟏 𝝈𝟐 −𝝈𝟐𝟐 𝟐𝒘𝜎𝑀
𝜕𝑤

• The optimal portfolio weight comes out as a parameter that depends on the marginal rate
of substitution between risk and return and the risk premium on the market portfolio relative
to its variance.

48
CARA utility and normal returns

• In the case where investors have CARA utility (constant absolute risk aversion) and
portfolio returns are normal, we can write all utility functions in the following way:
𝜌
𝑢 𝜇, 𝜎 2 = 𝜇 − 𝜎 2
2

• where 𝜌 is the risk aversion coefficient.

• We find here:
𝜕𝑢
𝜕𝑢 𝜕𝑢 𝜌 𝜕𝜎2 𝜌
=1; = − => 𝑀𝑅𝑆 = 𝜕𝑢 =−
𝜕𝜇 𝜕𝜎2 2 2
𝜕𝜇

• such that the MRS between risk and return is:

𝜕𝑤
𝜕𝜎 2 𝝆 𝒓 𝑭 − 𝑬 𝒓𝑴 𝑬 𝒓 𝑴 − 𝒓𝑭
𝑀𝑅𝑆 = − ⇒− = ⇒𝝆=
𝜕𝑤 𝟐 𝟐𝒘𝝈𝟐𝑴 𝒘𝝈𝟐𝑴
𝜕𝜇

49
CARA utility and normal returns

• The optimal weight on the market portfolio is given by:

𝑬 𝒓𝑴 −𝒓𝑭
𝒘=
𝝆𝝈𝟐𝑴

• The more risk averse the individual is, therefore, the smaller the
weight he puts on the market portfolio, as the risk aversion
coefficient enters into the denominator on the right hand side.

50
CARA utility and normal returns

𝑬 𝒓𝑴 − 𝒓𝑭
𝒘𝟐 =
𝝆𝟐 𝝈𝟐𝑴
Capital Market Line (CML)

𝑬 𝒓𝑴 − 𝒓 𝑭
𝒘𝟏 =
𝝆𝟏 𝝈𝟐𝑴

𝝈𝟏 𝝈𝟐
51
Alternative methods to find the weight on risky asset

• We have:
𝜇 = 𝑤𝑟𝑀 + (1 − 𝑤)𝑟𝐹 ;
2 2 2
𝜎𝑝2 = 𝑤 2 𝜎𝑀 + 1−𝑤 2 2
𝜎ด
𝑟𝐹 + 2𝑤 1 − 𝑤 𝜌𝑚,𝑟𝐹 𝜎𝑀 𝜎
ด𝑟𝐹 = 𝑤 𝜎𝑀 ;
=0 =0

• And
𝜌 2 𝜌
𝐸 𝑢 𝑥 = 𝜇 − 𝜎𝑝2 = 𝑤𝐸(𝑟𝑀 ) + 1 − 𝑤 𝑟𝐹 − 𝑤 2 𝜎𝑀
2 2
𝑑𝑢 2𝜌
= 𝐸(𝑟𝑀 ) − 𝑟𝐹 − 2𝑤𝜎𝑀 =0
𝑑𝑤 2
𝑬 𝒓𝑴 −𝒓𝑭
𝒘= or
𝝆𝝈𝟐𝑴

𝑬 𝒓𝑴 −𝒓𝑭
𝝆=
𝒘𝝈𝟐𝑴

52
Alternative methods to find the weight on risky asset

Example
• 𝑟𝐹 = 4%
• 𝑟𝑀 = 8%
• 𝑉𝑎𝑟 𝑟𝑀 = 16%
• If the value of Absolute Risk Aversion coefficient is 𝜌 = 0.8
• The optimal portfolio weight on the risky asset is
𝑬 𝒓𝑴 −𝒓𝑭 𝟖%−𝟒%
• 𝒘= = = 𝟎. 𝟑𝟏𝟐𝟓 𝒐𝒓 𝟑𝟏. 𝟐𝟓%
𝝆𝝈𝟐𝑴 (𝟎.𝟖)(𝟏𝟔%)

• 1 − 𝑤 = 68.75%

53
Alternative methods to find the weight on risky asset

• Find the value of 𝜌 such that the optimal portfolio weight on the risk-free asset is 80%
• 𝑤 = 20%

𝑬 𝒓𝑴 −𝒓𝑭 𝟖%−𝟒%
• 𝝆= = = 𝟏. 𝟐𝟓
𝒘𝝈𝟐𝑴 (𝟎.𝟐)(𝟏𝟔%)

54
Exercise 2

Assume a portfolio of two assets, asset 1 and asset 2:

What proportion of the total investment should be invested in each of the two assets to
minimize the variance of a two-asset portfolio when the assets have equal standard deviations
(𝜎1 = 𝜎2 )?

55
Exercise 2

• Let 𝑤 be the proportion of investment in asset 1.

• When 𝜎1 = 𝜎2 , we have:
𝜎 2 = 𝑤 2 𝜎12 + 2𝑤 1 − 𝑤 𝜎1 𝜎2 𝜌12 + 1 − 𝑤 2 𝜎22 … (𝑖𝑓𝜎1 ≠ 𝜎2 )
𝜎 2 = 𝑤 2 𝜎12 + 2𝑤 1 − 𝑤 𝜎12 𝜌12 + 1 − 𝑤 2 𝜎12 … (𝑖𝑓𝜎1 = 𝜎2 )
𝜎 2 = 𝜎12 𝑤 2 + 2𝑤𝜌12 − 2𝑤 2 𝜌12 + 1 − 2𝑤 + 𝑤 2
𝜎 2 = 𝜎12 2𝑤 2 + 2𝑤𝜌12 − 2𝑤 2 𝜌12 + 1 − 2𝑤
For a minimum variance:
𝑑𝜎 2
= 𝜎12 4𝑤 + 2𝜌12 − 4𝑤𝜌12 − 2 =0
𝑑𝑤
4𝑤 + 2𝜌12 − 4𝑤𝜌12 − 2=0 ⇒ 4𝑤 1 − 𝜌12 − 2 1 − 𝜌12 = 0
2 1−𝜌12 1
𝑤=4 1−𝜌12
=2

Therefore, if 𝜎1 = 𝜎2 , return variance of the portfolio will be minimised by choosing 𝑤 = 0.5

regardless of the correlation so long as the correlation 𝜌12 is not +1.

56
Exercise 2

• Let 𝑤 be the proportion of investment in asset 1.

• When 𝜎1 = 𝜎2 , we have:
𝜎 2 = 𝑤 2 𝜎12 + 2𝑤 1 − 𝑤 𝜎1 𝜎2 𝜌12 + 1 − 𝑤 2 𝜎22 … (𝑖𝑓𝜎1 ≠ 𝜎2 )
𝜎 2 = 𝑤 2 𝜎12 + 2𝑤 1 − 𝑤 𝜎12 𝜌12 + 1 − 𝑤 2 𝜎12 … (𝑖𝑓𝜎1 = 𝜎2 )
𝜎 2 = 𝜎12 𝑤 2 + 2𝑤𝜌12 − 2𝑤 2 𝜌12 + 1 − 2𝑤 + 𝑤 2
𝜎 2 = 𝜎12 2𝑤 2 + 2𝑤𝜌12 − 2𝑤 2 𝜌12 + 1 − 2𝑤
For a minimum variance:
𝑑𝜎 2
𝑑𝑤
= 𝜎12 4𝑤 + 2𝜌12 − 4𝑤𝜌12 − 2 =0

4𝑤 + 2𝜌12 − 4𝑤𝜌12 − 2=0 ⇒ 4𝑤 1 − 𝜌12 − 2 1 − 𝜌12 = 0

2 1−𝜌12 1
𝑤= =
4 1−𝜌12 2

Therefore, if 𝜎1 = 𝜎2 , return variance of the portfolio will be minimised by choosing 𝑤 = 0.5 regardless of the
correlation so long as the correlation 𝜌12 is not +1.

57
Exercise 3

Assume a portfolio of two assets, asset 1 and asset 2:

What proportion of the total investment should be invested in each of the two assets to
minimize the variance of a two-asset portfolio when the asset returns have a correlation of -1,
and standard deviations 𝜎1 and 𝜎2 , respectively?

58
Exercise 3

Let 𝑤 be the proportion of investment in asset 1.

𝜎 2 = 𝑤 2 𝜎12 + 2𝑤 1 − 𝑤 𝜎1 𝜎2 𝜌12 + 1 − 𝑤 2 𝜎22
𝜎 2 = 𝑤 2 𝜎12 − 2𝑤 1 − 𝑤 𝜎1 𝜎2 + 1 − 𝑤 2 𝜎22 … (𝜌12 = −1)
= 𝑤 2 𝜎12 − 2𝑤𝜎1 𝜎2 + 2𝑤 2 𝜎1 𝜎2 + 𝜎22 − 2𝑤𝜎22 + 𝑤 2 𝜎22
= 𝑤 2 𝜎12 + 2𝜎1 𝜎2 + 𝜎22 − 2𝑤 𝜎22 + 𝜎1 𝜎2 + 𝜎22
= 𝑤 2 𝜎1 + 𝜎2 2 − 2𝑤𝜎2 𝜎1 + 𝜎2 + 𝜎22
𝜎 2 = 𝑤 𝜎1 + 𝜎2 − 𝜎2 2

• Zero-variance portfolio is achieved when:

• 𝑤 𝜎1 + 𝜎2 − 𝜎2 = 0
• Therefore,
𝜎2 𝜎1
𝑤= and 1 − 𝑤 =
𝜎1 +𝜎2 𝜎1 +𝜎2
59
Exercise 3

Let 𝑤 be the proportion of investment in asset 1.

𝜎 2 = 𝑤 2 𝜎12 + 2𝑤 1 − 𝑤 𝜎1 𝜎2 𝜌12 + 1 − 𝑤 2 𝜎22
𝜎 2 = 𝑤 2 𝜎12 − 2𝑤 1 − 𝑤 𝜎1 𝜎2 + 1 − 𝑤 2 𝜎22 … (𝜌12 = −1)
= 𝑤 2 𝜎12 − 2𝑤𝜎1 𝜎2 + 2𝑤 2 𝜎1 𝜎2 + 𝜎22 − 2𝑤𝜎22 + 𝑤 2 𝜎22
= 𝑤 2 𝜎12 + 2𝜎1 𝜎2 + 𝜎22 − 2𝑤 𝜎22 + 𝜎1 𝜎2 + 𝜎22
= 𝑤 2 𝜎1 + 𝜎2 2 − 2𝑤𝜎2 𝜎1 + 𝜎2 + 𝜎22
𝜎 2 = 𝑤 𝜎1 + 𝜎2 − 𝜎2 2

• Zero-variance portfolio is achieved when:

• 𝑤 𝜎1 + 𝜎2 − 𝜎2 = 0
• Therefore,
𝜎2 𝜎1
𝑤= and 1 − 𝑤 =
𝜎1 +𝜎2 𝜎1 +𝜎2
60