Investment Analysis and

Portfolio Management
Lecture 3
Gareth Myles
FT 100 Index
£ and $
Risk
 Variance
 The standard measure of risk is the
variance of return
or
 Its square root: the standard deviation

 Sample variance
 The value obtained from past data
 Population variance
 The value from the true model of the data
Sample Variance

General Motors Stock Price 1962-2008

Sample Variance

Year 93-94 94-95 95-96 96-97 97-98

Return % 36.0 -9.2 17.6 7.2 34.1
Year 98-99 99-00 00-01 01-02 02-03
Return % -1.2 25.3 -16.6 12.7 -40.9

Return on General Motors Stock 1993-2003

Sample Variance
40
30
20
10
0
-10 93- 94- 95- 96- 97- 98- 99- 00- 01- 02-
94 95 96 97 98 99 00 01 02 03
-20
-30
-40
-50
Graph of return
Sample Variance
 With T observations sample variance is
1 T
    rt  r  2
2
2 0
T t 1
 The standard deviation is
1 T
  rt  r 
2
  0
T t 1
 Both these are biased estimators
 The unbiased estimators are
1 T 1 T
 T2 1    rt  r 
2
  rt  r 
2
 T 1 
T  1 t 1 T  1 t 1
Sample Variance
 For the returns on the General Motors stock,
the mean return is 6.5
 Using this value, the deviations from the
mean and their squares are given by
Year 93-94 94-95 95-96 96-97 97-98
rt  r 29.5 -15.7 11.1 0.7 27.6
 rt  r  2 870.25 246.49 123.21 0.49 761.76
Year 98-99 99-00 00-01 01-02 02-03
rt  r -7.7 18.8 -23.1 6.2 -47.4
 rt  r  2 59.29 353.44 533.61 38.44 2246.76
Sample Variance
 After summing and averaging, the variance is
 2  523.4
 The standard deviation is

  523.4  22.88
 This information can be used to compare
different securities
 A security has a mean return and a variance of
the return
Sample Covariance
 The covariance measures the way the returns
on two assets vary relative to each other
 Positive: the returns on the assets tend to rise and fall
together
 Negative: the returns tend to change in opposite
directions
 Covariance has important consequences for
portfolios
Asset Return in 2001 Return in 2002
A 10 2
B 2 10
Sample Covariance
 Mean return on each stock = 6
2 2
 Variances of the returns are  A   B  16
 Portfolio: 1/2 of asset A and 1/2 of asset B
1 1
 Return in 2001: rp  10  2  6
2 2
1 1
 Return in 2002: rp  2  10  6
2 2
 Variance of return on portfolio is 0
Sample Covariance
 The covariance of the return is
1 T
 AB    rAt  rA  rBt  rB 
T t 1
 It is always true that

 i.  AB   BA
2
 
ii. ii   i
Sample Covariance
 Example. The table provides the returns on three
assets over three years

Year 1 Year 2 Year 3

A 10 12 11
B 10 14 12
C 12 6 9

 Mean returns rA  11, rB  12, rC  9

Sample Covariance
 Covariance between A and B is
1
 AB   10  1110  12  12  1114  12  11  1112  12 
3
 1.333

 Covariance between A and C is

1
 AC   10  1112  9  12  11 6  9  11  11 9  9 
3
 2
Variance-Covariance Matrix

 Covariance between B and C is

1
 BC   10  1212  9  14  12 6  9  12  12 9  9 
3
 4
 The matrix is symmetric
A B C
A  A2
B  AB  B2
C  AC  BC  C2
Variance-Covariance Matrix
 For the example the variance-covariance
matrix is

A B C
A 0.666
B 1.333 2.66
C 2 4 6
Population Return and Variance

 Expectations: assign probabilities to outcomes

 Rolling a dice: any integer between 1 and 6 with
probability 1/6
 Outcomes and probabilities are:
{1,1/6}, {2,1/6}, {3,1/6}, {4,1/6}, {5,1/6}, {6,1/6}
 Expected value: average outcome if experiment
repeated
1 1 1 1 1 1
E[ X ]  1  2  3  4  5  6
6 6 6 6 6 6
 3.5
Population Return and Variance
 Formally: M possible outcomes
 Outcome j is a value xj with probability j
 Expected value of the random variable X is
M
E[ X ]    j x j
j 1
 The sample mean is the best estimate of the
expected value
Population Return and Variance

 After market analysis of Esso an analyst

determines possible returns in 2010

Return 2 6 9 12
Probability 0.2 0.3 0.3 0.2

 The expected return on Esso stock using this

data is
E[rEsso] = .2(2) + .3(6) + .3(9) + .2(12)
= 7.3
Population Return and Variance

 The expectation can be applied to functions of X

 For the dice example applied to X2
2 1 1 1 1 1 1
E[ X ]  1  4  9  16  25  36
6 6 6 6 6 6
 15.167
 And to X3
3 1 1 1 1 1 1
E[ X ]  1  8  27  64  125  216
6 6 6 6 6 6
 73.5
Population Return and Variance

 The expected value of the square of the

deviation from the mean is
1 1 1
E[ X  E[ X ] ]  1  3.5   2  3.5   3  3.5 2
2 2 2
6 6 6
1 1 1
  4  3.5   5  3.5   6  3.5
2 2 2
6 6 6
 2.9167
 This is the population variance
Modelling Returns
 States of the world
 Provide a summary of the information about
future return on an asset
 A way of modelling the randomness in asset
returns
 Not intended as a practical description
Modelling Returns
 Let there be M states of the world
 Return on an asset in state j is rj
 Probability of state j occurring is j
 Expected return on asset i is

E[r ]   1r1  ...   M rM

M
  j rj
j 1
Modelling Returns
 Example: The temperature next year may be
hot, warm or cold
 The return on stock in a food production
company in each state
State Hot Warm Cold
Return 10 12 18
 If each states occurs with probability 1/3, the
expected return on the stock is
1 1 1
E[r ]  10  12  18  13.333
3 3 3
Portfolio Expected Return
 N assets
 M states of the world
 Return on asset i in state j is rij
 Probability of state j occurring is j
 Xi proportion of the portfolio in asset i
 Return on the portfolio in state j
N
rPj   X i rij
i 1
Portfolio Expected Return
 The expected return on the portfolio
E  rP   1rP1  ...   M rPM
 Using returns on individual assets
N N
E  rP    1 X i ri1  ...    M X i riM
i 1 i 1
 Collecting terms this is
N
E  rP    X i  1ri1  ...   M riM 
i 1
 So N
rP   X i ri
i 1
Portfolio Expected Return
 Example: Portfolio of asset A (20%), asset
B (80%)
 Returns in the 5 possible states and
probabilities are:
State 1 2 3 4 5
Probability 0.1 0.2 0.4 0.1 0.2
Return on A 2 6 9 1 2
Return on B 5 1 0 4 3
Portfolio Expected Return
 For the two assets the expected returns are
rA  0.1  2  0.2  6  0.4  9  0.1  1  0.2  2  5.5
rB  0.1  5  0.2  1  0.4  0  0.1 4  0.2  3  1.7

 For the portfolio the expected return is

rP  0.2  5.5  0.8  1.7  2.46

Population Variance and
Covariance
 Population variance


 i  E  ri  E  ri  
2 2

 The sample variance is an estimate of this
 Population covariance
 ij  E  ri  E  ri    r j  E  r j   
 The sample covariance is an estimate of this
Population Variance and
Covariance
 M states of the world, return in state j is rij
 Probability of state j is j
 Population variance is
2
i
M

j 1

   j  rij  ri  2

 Population standard deviation is
M
 i    j rij  ri 2
j 1
  
Population Variance and
Covariance
 Example: The table details returns in five
possible states and the probabilities

State 1 2 3 4 5
Return 5 2 -1 6 3
Probability 0.1 0.2 0.4 0.1 0.2
 The population variance is
 2  .1 5  3 2  .2 2  3 2  .4  1  3 2  .1 6  3 2  .2 3  3 2
 7.9
Portfolio Variance
 Two assets A and B
 Proportions XA and XB
 Return on the portfolio rP
 Mean return rP
 Portfolio variance


 P2  E  rP  E  rP   2 
Portfolio Variance
 Population covariance between A and B is

 AB  E   rA  E  rA   rB  E  rB   

 For M states with probabilities j

 AB    j  rAj  rA  rBj  rB 
M

j 1
Portfolio Variance
 The portfolio return is
rP  X ArA  X B rB rP  X ArA  X B rB

 So
2
P 
 E  X ArA  X B rB  X ArA  X B rB  2

 Collecting terms

 P2  E  X A  rA  rA   X B  rB  rB   2 
Portfolio Variance
 Squaring


 P2  E X A 2  rA  rA  2  X B 2  rB  rB  2  2 X A X B  rA  rA  rB  rB  
 Separate the expectations

  
 P2  X A 2 E  rA  rA  2  X B 2 E  rB  rB  2 
 2 X A X B E   rA  rA  rB  rB  
 Hence
 P2  X A2 A2  X B 2 B2  2 X A X B AB
Portfolio Variance
 Example: Portfolio consisting of
 1/3 asset A
 2/3 asset B
 The variances/covariance are
2 2
 A  2.333,  B  8.333,  AB  3.333
 The portfolio variance is
2 2
1  2  1  2 
 P2    2.333    8.333  2  3.333
 3  3  3  3 
 9.148
Correlation Coefficient
 The correlation coefficient is defined by
 AB
 AB 
 A B
 Value satisfies
 1   AB  1
  AB  1 perfect positive correlation
rB

rA
Correlation Coefficient
  AB  1 perfect negative correlation
rB

rA

 Variance of the return of a portfolio

2 2 2 2 2
 P  X A A  X B B  2 X A X B  AB A B
Correlation Coefficient
 Example: Portfolio consisting of
 1/4 asset A
 3/4 asset B
 The variances/correlation are
 A2  16,  B2  9,  AB  0.5
 The portfolio variance is
2 2
2 1 3  1  3 
 P    16    9  2  (0.5)(4)(3)
4 4  4  4 
 3.8125
General Formula
 N assets, proportions Xi
 Portfolio variance is

2
N  2 2 N 
P    X i  i   X i X k  ik 
i 1 k 1, k  i 
2
 
But ii   i so
N N
2
P    X i X k  ik
i 1k 1
Effect of Diversification
 Diversification: a means of reducing risk
 Consider holding N assets
 Proportions Xi = 1/N
 Variance of portfolio

N  1  2 N 1
2 
2 2
 P       i      ik 
i 1  N  k 1, k  i  N  
 
Effect of Diversification
 N terms in the first summation, N[ N-1] in
the second
 Gives 2  1  N  1  2  N  1 N N  1 
 P       i        ik
 N i 1 N   N i 1k 1  N  1 N 
k i
 Define
N  1 2 N N  1 
 a2     i  ab      ik
i 1 N  i 1k 1  N  1 N 
k i
 Then  1  2  N  1
 P2    a    ab
N   N 
Effect of Diversification
 Let N tend to infinity (extreme diversification)
 Then
1 2  N 1
 N  a  0  N  ab   ab
 Hence
2
 P   ab
 In a well-diversified portfolio only the covariance
between assets counts for portfolio variance