Professional Documents
Culture Documents
Modern Portfolio Theory and Investment Analysis 6th Solutions PDF
Modern Portfolio Theory and Investment Analysis 6th Solutions PDF
Chapter 1: Problem 1
A. Opportunity Set
With one dollar, you can buy 500 red hots and no rock candies (point A),
or 100 rock candies and no red hots (point B), or any combination of red hots
and rock candies (any point along the opportunity set line AB).
0.2 X 1Y 100
That is, the money spent on candies, where red hots sell for 0.2 cents a piece
and rock candy sells for 1 cent a piece, cannot exceed 100 cents ($1.00).
Solving the above equation for X gives:
X 500 5Y
which is the equation of a straight line, with an intercept of 500 and a slope of 5.
Below is one indifference map. The indifference curves up and to the right
indicate greater happiness, since these curves indicate more consumption from
both candies. Each curve is negatively sloped, indicating a preference of more
to less, and each curve is convex, indicating that the rate of exchange of red
hots for rock candies decreases as more and more rock candies are consumed.
Note that the exact slopes of the indifference curves in the indifference map will
depend an the individuals utility function and may differ among students.
Chapter 1: Problem 2
A. Opportunity Set
or
C2 = $42 1.1C1
which is the equation of a straight line with an intercept of $42 and a slope of
1.1.
B. Indifference Map
C1C 2
U(C1 , C 2 ) 1 C1 C 2
50
C. Solution
The optimum solution is where the opportunity set is tangent to the highest
possible indifference curve (the point labeled E in the following graph).
If you consume nothing at time 1 and instead invest all of your time-1
income at a riskless rate of 10%, then at time 2 you will be able to consume all of
your time-2 income plus the proceeds earned from your investment:
If you consume nothing at time 2 and instead borrow at time 1 the present value
of your time-2 income at a riskless rate of 10%, then at time 1 you will be able to
consume all of the borrowed proceeds plus your time-1 income:
Chapter 1: Problem 4
If you consume nothing at time 1 and instead invest all of your wealth plus your
time-1 income at a riskless rate of 5%, then at time 2 you will be able to consume
all of your time-2 income plus the proceeds earned from your investment:
= $93,500 1.05C1
Chapter 1: Problem 5
With Job 1 you can consume $30 + $50 (1.05) = $82.50 at time 2 and
nothing at time 1, $50 + $30 y (1.05) = $78.60 at time 1 and nothing at time 2, or
any consumption pattern of time 1 and time 2 consumption shown along the line
AB: C2 = $82.50 1.05C1.
With Job 2 you can consume $40 + $40 (1.05) = $82.00 at time 2 and
nothing at time 1, $40 + $40 y (1.05) = $78.10 at time 1 and nothing at time 2, or
any consumption pattern of time 1 and time 2 consumption shown along the line
CD: C2 = $82.00 1.05C1.
The individual should select Job 1, since the opportunity set associated
with it (line AB) dominates the opportunity set of Job 2 (line CD).
With an interest rate of 10% and income at both time 1 and time 2 of
$5,000, the opportunity set is given by the line AB:
With an interest rate of 20% and income at both time 1 and time 2 of $5,000, the
opportunity set is given by the line CD:
Chapter 1: Problem 7
50 C1
For P = 50, this is simply a plot of the function C 2 .
1 C1
100 C1
For P = 100, this is simply a plot of the function C 2 .
1 C1
Chapter 1: Problem 9
Let X = the number of pizza slices, and let Y = the number of hamburgers.
Then, if pizza slices are $2 each, hamburgers are $2.50 each, and you have $10,
your opportunity set is given algebraically by
$2X + $2.50Y = $10
Solving the above equation for X gives X = 5 1.25Y, which is the equation for a
straight line with an intercept of 5 and a slope of 1.25.
Assuming you like both pizza and hamburgers, your indifference curves will
be negatively sloped, and you will be better off on an indifference curve to the
right of another indifference curve. Assuming diminishing marginal rate of
substitution between pizza slices and hamburgers (the lower the number of
hamburgers you have, the more pizza slices you need to give up one more burger
without changing your level of satisfaction), your indifference curves will also be
convex.
Chapter 1: Problem 10
If you consume C1 at time 1 and invest (lend) the rest of your time-1 income at 5%,
your time-2 consumption (C2) will be $50 from your time-2 income plus ($50
C1)(1.05) from your investment. Algebraically, the opportunity set is thus
If C1 is 0 (no time-1 consumption), then from the above equation C2 will be $102.50.
If C2 is 0, then C1 will be $97.62. Graphically, the opportunity set appears below,
along with a typical family of indifference curves.
If you consume C1 at time 1 and invest (lend) the rest of your time-1 income
at 20%, your time-2 consumption (C2) will be $10,000 from your time-2 income plus
$10,000 from your inheritance plus ($10,000 - C1)(1.20) from your investment. The
opportunity set is thus
If C2 is 0 (no time-2 consumption), then you can borrow the present value of your
time-2 income and your time-2 inheritance and spend that amount along with
your time-1 income on time-1 consumption. Solving the above equation for C1
when C2 is 0 gives C1 = $26,666.67, which is the maximum that can be consumed
at time 1. Similarly, if C1 is 0 (no time-1 consumption), then you can invest all of your
time-1 income at 20% and spend the future value of your time-1 income plus your
time-2 income and inheritance on time-2 consumption. From the above equation,
C2 will be $32,000 when C1 is 0, which is the maximum that can be consumed at
time 2.
Chapter 1: Problem 12
If you consume nothing at time 2, then you can borrow the present value of
your time-2 income for consumption at time 1. If the borrowing rate is 10% and
your time-2 income is $100, then the present value (at time 1) of your time-2
income is $100/(1.1) = $90.91. You can borrow this amount and spend it along with
your time- income of $100 on time-1 consumption. So the maximum you can
consume at time 1 is $90.91 + $100 = $190.91. If you consume nothing at time 1
and instead invest all of your time-1 income of $100 at the lending rate of 5%, the
future value (in period 2) of your period 1 income will be $100(1.05) = $105. You
can then spend that amount along with your time 2 income of $100 on time-2
consumption. So the maximum you can consume at time 2 is $105 + $100 = $205.
With two different interest rates, we have two separate equations for opportunity
sets: one for borrowers and one for lenders.
If you only consume some of your time 1 income at time 1 and invest the
rest at 5%, you have the following opportunity set: C2 = 100 + (100 - C1)(1.05) = 205 -
1.05C1. If you only consume some of your time-2 income at time 2 and borrow the
present value of the rest at 10% for consumption at time 1, your opportunity set is:
C1 = 100 + (100 - C2)/(1.1) = 190.91 - C2/1.1,or, solving the equation for C2,
C2 = 210 - 1.1C1
The lines intersect at point E (which represents your income endowment for times 1
and 2). Moving along the lines above point E represents lending (investing some
time-1 income); moving along the lines below point E represents borrowing
(spending more than your time-1 income on time-1 consumption). Since you can
only lend at 5%, line segment AE represents your opportunity set if you choose to
lend. Since you must borrow at 10%, line segment ED represents your opportunity
set if you choose to borrow. So your total opportunity set is represented by AED.
Chapter 4: Problem 1
Standard deviation of return is the square root of the sum of the squares of
each outcome minus the mean times the associated probability.
1 2 3 4
1 8 4 12 0
2 4 2 6 0
3 12 6 18 0
4 0 0 0 10.7
4
Correlation of return between Assets 1 and 2 = U 12 1.
2.83 u 1.41
1 2 3 4
1 1 1 1 0
2 1 1 1 0
3 1 1 1 0
4 0 0 0 1
Elton, Gruber, Brown and Goetzmann 13
Modern Portfolio Theory and Investment Analysis, 6th Edition
Solutions To Text Problems: Chapter 4
C. Portfolio Expected Return
B 13%
C 12%
D 10%
E 13%
G 12.67%
H 12.67%
Portfolio Variance
B 12.5
C 4.6
D 2
E 7
G 2
H 6.7
A. Monthly Returns
Month
Security 1 2 3 4 5 6
RA
3.7% 0.4% 6.5% 1.4% 6.2% 2.1% 1.22%
6
RB 2.95%
RC 7.92%
VA
3.7% 1.22% 2 0.4% 1.22% 2 6.5% 1.22% 2 1.4% 1.22% 2 6.2% 1.22% 2 2.1% 1.22% 2
6
15.34 3.92%
VB 14.42 3.8%
VC 46.02 6.78%
3.7% 1.22% u 10.5% 2.95% 0.4% 1.22% u 0.5% 2.95% 6.5% 1.22% u 3.7% 2.95%
V AB 1.4% 1.22% u 1.0% 2.95% 6.2% 1.22% u 3.4% 2.95% 2.1% 1.22% u 1.4% 2.95%
6
2.17
V AC 7.24 ; V BC 19.89
2.17
U AB 0.15
3.92 u 3.8
U AC 0.27 ; U BC 0.77
RP 2 4.57%
V P2 18.96 4.35%
RP 3 5.44%
V P3 5.17 2.27%
RP 4 4.03%
V P4 6.09 2.47%
It is shown in the text below Table 4.8 that a formula for the variance of an
equally weighted portfolio (where Xi = 1/N for i = 1, , N securities) is
V P = 1/N V j V kj V kj
2 2
2
Portfolio Size (N) VP
5 18
10 14
20 12
50 10.8
100 10.4
Chapter 4: Problem 4
V P2 1/ N u 50 10 10 11
As shown in the text, if the portfolio contains only one security, then the portfolios
average variance is equal to the average variance across all securities, V 2j . If
instead an equally weighted portfolio contains a very large number of securities,
then its variance will be approximately equal to the average covariance of the
pairs of securities in the portfolio, V kj . Therefore, the fraction of risk that of an
individual security that can be eliminated by holding a large portfolio is expressed
by the following ratio:
V i2 V kj
V i2
From Table 4.9, the above ratio is equal to 0.6 (60%) for Italian securities and 0.8
(80%) for Belgian securities. Setting the above ratio equal to those values and
solving for V kj gives V kj 0.4V i2 for Italian securities and V kj 0.2V i2 for Belgian
securities.
V i2 V kj V i2 0.4V i2
Thus, the ratio equals 1.5 for Italian securities and
V kj 0.4V i2
V i2 0.2V i2
4 for Belgian securities.
0.2V i2
If the average variance of a single security, V 2j , in each country equals 50, then
Using the formula shown in the preceding answer to Problem 3 with V 2j = 50 and
either V kj = 20 for Italy or V kj = 10 for Belgium we have:
5 26 18
20 21.5 12
The formula for an equally weighted portfolio's variance that appears below Table
4.8 in the text is
V P = 1/N V j V kj V kj
2 2
.942N = 39.561
N = 41.997.
Chapter 5: Problem 1
V 12 = 4 V 13 = 12 V 14 = 0 V 23 = 6 V 24 = 0 V 34 = 0
U 12 = 1 U 13 = 1 U 14 = 0 U 23 = 1.0 U 24 = 0 U 34 = 0
In this problem, we will examine 2-asset portfolios consisting of the following pairs
of securities:
Pair Securities
A 1 and 2
B 1 and 3
C 1 and 4
D 2 and 3
E 2 and 4
F 3 and 4
A.1
We want to find the weights, the standard deviation and the expected return of
the minimum-risk porfolio, also known as the global minimum variance (GMV)
portfolio, of a pair of assets when short sales are not allowed.
V j2 V ij
X iGMV
V i2 V j2 2V ij
X GMV
j 1 X iGMV
V 22 V 12 2 (4) 6 1
X1GMV (or 33.33%)
V 12 V 22 2V 12 8 2 (2)(4) 18 3
1 2
X 2GMV 1 X1GMV 1 (or 66.67%)
3 3
This in turn gives the following for the GMV portfolio of Pair A:
1 2
R GMV u 12% u 6% 8%
3 3
2 2
1 2 1 2
V GMV
2
8 2 2 4 0
3 3 3 3
V GMV 0
Recalling that U 12 = 1, the above result demonstrates the fact that, when two
assets are perfectly negatively correlated, the minimum-risk portfolio of those two
assets will have zero risk.
This means that the GMV portfolio of assets 1 and 3 involves short selling asset 3.
But if short sales are not allowed, as is the case in this part of Problem 1, then the
GMV portfolio involves placing all of your funds in the lower risk security (asset 1)
and none in the higher risk security (asset 3). This is obvious since, because the
correlation between assets 1 and 3 is +1.0, portfolio risk is simply a linear
combination of the risks of the two assets, and the lowest value that can be
obtained is the risk of asset 1.
D (i = 2, j = 3) 0.75 0.25 8% 0%
Pair A
Pair C
Pair E
The efficient set is the positively sloped part of the curve, starting at the GMV
portfolio and ending at security 4.
The efficient set is the positively sloped part of the curve, starting at the GMV
portfolio and ending at security 3.
B.1
When short selling is allowed, all of the GMV portfolios shown in Part A.1 above
are the same except the one for Pair B (assets 1 and 3). In the no-short-sales case
in Part A.1, the GMV portfolio for Pair B was the lower risk asset 1 alone.
However, applying the GMV weight formula to Pair B yielded the following
weights:
X1GMV 3 (300%) and X 3GMV 2 (200%)
This means that the GMV portfolio of assets 1 and 3 involves short selling asset 3 in
an amount equal to twice the investors original wealth and then placing the
original wealth plus the proceeds from the short sale into asset 1. This yields the
following for Pair B when short sales are allowed:
R GMV 3 u 12% 2 u 14% 8%
V GMV
2
32 8 22 18 23 212 0
V GMV 0
Recalling that U 13 = +1, this demonstrates the fact that, when two assets are
perfectly positively correlated and short sales are allowed, the GMV portfolio of
those two assets will have zero risk.
Pair A
The efficient set is the positively sloped line segment through security 1 and out
toward infinity.
Pair B
Pair D
The efficient set is the positively sloped line segment through security 3 and out
toward infinity.
The efficient set is the positively sloped part of the curve, starting at the GMV
portfolio and extending past security 4 toward infinity.
Pair F
The efficient set is the positively sloped part of the curve, starting at the GMV
portfolio and extending past security 3 toward infinity.
Chapter 5: Problem 2
In this problem, we will examine 2-asset portfolios consisting of the following pairs
of securities:
Pair Securities
1 A and B
2 A and C
3 B and C
A.1
We want to find the weights, the standard deviation and the expected return of
the minimum-risk porfolio, also known as the global minimum variance (GMV)
portfolio, of a pair of assets when short sales are not allowed.
We further know that the compostion of the GMV portfolio of any two assets i
and j is:
V j2 V ij
X iGMV
V i2 V j2 2V ij
X GMV
j 1 X iGMV
Applying the above GMV weight formula to Pair 1 yields the following weights:
V B2 V AB 14.42 2.17
X AGMV 0.482 (or 48.2%)
V 2
A V 2
B 2V AB 15.34 14.42 (2)(2.17)
This in turn gives the following for the GMV portfolio of Pair 1:
V GMV
2
0.4822 15.34 0.5182 14.42 20.4820.5182.17 8.52
V GMV 2.92%
Pair 1
The efficient set is the positively sloped part of the curve, starting at the GMV
portfolio and ending at security B.
The efficient set is the positively sloped part of the curve, starting at the GMV
portfolio and ending at security C.
Pair 3
The efficient set is the positively sloped part of the curve, starting at the GMV
portfolio and ending at security C.
B.1
When short selling is allowed, all of the GMV portfolios shown in Part A.1 above
remain the same.
Pair 1
The efficient set is the positively sloped part of the curve, starting at the GMV
portfolio and extending past security B toward infinity.
The efficient set is the positively sloped part of the curve, starting at the GMV
portfolio and extending past security C toward infinity.
Pair 3
The efficient set is the positively sloped part of the curve, starting at the GMV
portfolio and extending past security C toward infinity.
Chapter 5: Problem 3
The answers to this problem are given in the answers to part A.1 of Problem 2.
Chapter 5: Problem 4
RP = X C RC + 1 X C RS
+ +
= V P V S RC + 1 V P V S RS
VC +V S V C +V S
= RS + V P RC V S RC V P RS V S RS
+
VC +V S
= RS + RC RS V S RC RS V P .
V C + V S V C + V S
The above equation is that of a straight line in expected return standard deviation
space, with an intercept equal to the first term in brackets and a slope equal to
the second term in brackets.
Substitute the above expression for XC into the equation for expected return and
simplify:
RP = X C RC + 1 X C RS
= V S V P RC + 1 V S V P RS
V C +V S V C +V S
= RS + V S RC V P RC V S RS V P RS
+
VC VS
+
= RS + RC RS V S RS RC V P .
V C + V S V C + V S
The above equation is also that of a straight line in expected return standard
deviation space, with an intercept equal to the first term in brackets and a slope
equal to the second term in brackets. The intercept term for the above equation is
identical to the intercept term for the first derived equation. The slope term is
equal to 1 times the slope term of the first derived equation. So when one
equation has a positive slope, the other equation has a negative slope (when the
expected returns of the two assets are equal, the two lines are coincident), and
both lines meet at the same intercept.
Chapter 5: Problem 5
When equals 1, the least risky "combination" of securities 1 and 2 is security 2 held
alone (assuming no short sales). This requires X1 = 0 and X2 = 1, where the X's are
the investment weights. The standard deviation of this "combination" is equal to
the standard deviation of security 2; P = 2 = 2.
When equals -1, we saw in Chapter 5 that we can always find a combination of
the two securities that will completely eliminate risk, and we saw that this
combination can be found by solving X1 = 2/(1 + 2). So, X1 = 2/(5 + 2) = 2/7, and
since the investment weights must sum to 1, X2 = 1 - X1 = 1 - 2/7 = 5/7. So a
combination of 2/7 invested in security 1 and 5/7 invested in security 2 will
completely eliminate risk when equals -1, and P will equal 0.
V P = X1 V 1 1 X1 V 2
2 2 2 2
2 2
4 25
VP u 25 u 4
29 29
400 2500
841 841
2900
841
1.86%
Chapter 5: Problem 6
If the riskless rate is 10%, then the risk-free asset dominates both risky assets in terms
of risk and return, since it offers as much or higher expected return than either risky
asset does, for zero risk. Assuming the investor prefers more to less and is risk averse,
the optimal investment is the risk-free asset.
Chapter 6: Problem 1
Z1 = 0.292831
Z2 = 0.009118
Z3 = 0.003309
This results in the following set of weights for the optimum (tangent) portfolio:
X1 = .95929 (95.929%)
X2 = .02987 (2.987%)
X3 = .01084 (1.084%
The optimum portfolio has a mean return of 10.146% and a standard deviation of
4.106%.
Chapter 6: Problem 2
RF = 6% RF = 8% RF = 10%
Z1 3.510067 1.852348 0.194631
Z2 1.043624 0.526845 0.010070
Z3 0.348993 0.214765 0.080537
Chapter 6: Problem 3
Since short sales are not allowed, this problem must be solved as a quadratic
programming problem. The formulation of the problem is:
RP RF
max T
X VP
subject to:
X
i 1
i 1
Xi t 0 i
This problem is most easily solved using The Investment Portfolio software that
comes with the text, but, since all pairs of assets are assumed to have the same
correlation coefficient of 0.5, the problem can also be solved manually using the
constant correlation form of the Elton, Gruber and Padberg Simple Techniques
described in a later chapter.
To use the software, open up the Markowitz module, select file then new then
group constant correlation to open up a constant correlation table. Enter the
input data into the appropriate cells by first double clicking on the cell to make it
active. Once the input data have been entered, click on optimizer and then
run optimizer (or simply click on the optimizer icon). At that point, you can either
select full Markowitz or simple method.
If you select full Markowitz, you then select short sales allowed/riskless lending
and borrowing and then enter 4 for both the lending and borrowing rate and
click OK. A graph of the efficient frontier then appears. You may then hit the
Tab key to jump to the tangent portfolio, then click on optimizer and then
show portfolio (or simply click on the show portfolio icon) to view and print the
composition (investment weights), mean return and standard deviation of the
tangent (optimum) portfolio.
If instead you select simple method, you then select short sales allowed with
riskless asset and enter 4 for the riskless rate and click OK. A table showing the
investment weights of the tangent portfolio then appears.
Regardless of the method used, the resulting investment weights for the optimum
portfolio are as follows:
Asset i Xi
1 5.999%
2 17.966%
3 21.676%
4 0.478%
5 29.585%
6 12.693%
7 59.170%
8 14.793%
9 3.442%
10 189.224%
Chapter 6: Problem 5
Since the given portfolios, A and B, are on the efficient frontier, the GMV portfolio
can be obtained by finding the minimum-risk combination of the two portfolios:
V B2 V AB 16 20 1
X AGMV
V 2
A V 2
B 2V AB 36 16 2 u 20 3
1
X BGMV 1 X AGMV 1
3
Also, since the two portfolios are on the efficient frontier, the entire efficient frontier
can then be traced by using various combinations of the two portfolios, starting
with the GMV portfolio and moving up along the efficient frontier (increasing the
weight in portfolio A and decreasing the weight in portfolio B). Since XB = 1 XA
the efficient frontier equations are:
RP X A R A 1 X A R B 10 X A 8 u 1 X A
X A2 V A2 1 X A V B2 2 X A 1 X A V AB
2
VP
36X A2 161 X A 40 X A 1 X A
2
Chapter 7: Problem 1
We will illustrate the answers for stock A and the market portfolio (S&P 500); the
answers for stocks B and C are found in an identical manner.
R
t 1
At
RA
12
12.05 15.27 4.12 1.57 3.16 2.79 8.97 1.18 1.07 12.75 7.48 0.94
12
2.946%
The sample mean monthly return on the market portfolio (the answer to part 1.E) is:
12
R
t 1
mt
Rm
12
12.28 5.99 2.41 4.48 4.41 4.43 6.77 2.11 3.46 6.16 2.47 1.15
12
3.005%
Using data given in the problem and the above two sample mean monthly returns,
we have the following:
Month t RAt RA R At RA 2
Rmt Rm R
mt Rm
2
R
At
RA Rmt Rm
1 9.104 82.883 9.275 86.026 84.44
2 12.324 151.881 2.985 8.910 36.79
3 -7.066 49.928 -0.595 0.354 4.2
4 -1.376 1.893 1.475 2.176 -2.03
5 0.214 0.046 1.405 1.974 0.3
6 -5.736 32.902 1.425 2.031 -8.17
7 -11.916 141.991 -9.775 95.551 116.48
8 -4.126 17.024 -5.115 26.163 21.1
9 -1.876 3.519 0.455 0.207 -0.85
10 9.804 96.118 3.155 9.954 30.93
11 4.534 20.557 -0.535 0.286 -2.43
12 -3.886 15.101 -4.155 17.264 16.15
R
12
2
At RA
613.84
V A2 t 1
51.15
12 12
VA 51.15 7.15%
The sample variance (the answer to part 1.F) and standard deviation of the
market portfolios monthly return are:
R
12
2
mt Rm
250.90
Vm
2 t 1
20.91
12 12
Vm 20.91 4.57%
The sample covariance of the returns on stock A and the market portfolio is:
>R @
12
At RA Rmt Rm
296.91
V Am t 1
24.74
12 12
The sample correlation coefficient of the returns on stock A and the market
portfolio (the answer to part 1.D) is:
V Am 24.74
U Am 0.757
V AV m 7.15 u 4.57
V Am 24.74
EA 1.183
Vm 2
20.91
Each months sample residual is security As actual return that month minus the
return that month predicted by the regression. The regressions predicted monthly
return is:
Since the sample residuals sum to 0 (because of the way the sample alpha and
beta are calculated), the sample mean of the sample residuals also equals 0 and
the sample variance and standard deviation of the sample residuals (the answer
to part 1.C) are:
H H
12 12
At HA At
262.36
V H2A t 1 t 1
21.863
12 12 12
V HA 21.863 4.676%
correlation
with market 0.757 0.684 0.652
standard deviation
of sample residuals* 4.676% 4.983% 12.341%
*Note that most regression programs use N 2 for the denominator in the sample
residual variance formula and use N 1 for the denominator in the other variance
formulas (where N is the number of time series observations). As is explained in the
text, we have instead used N for the denominator in all the variance formulas. To
convert the variance from a regression program to our results, simply multiply the
N2 N 1
variance by either or .
N N
Chapter 7: Problem 2
A.
A.1
The Sharpe single-index model's formula for a security's mean return is
Ri = D i + E i R m
Using the alpha and beta for stock A along with the mean return on the market
portfolio from Problem 1 we have:
Similarly:
RB 6.032% ; RC 3.556%
V i2 E i2 V m
2
V H2i
Using the beta and residual standard deviation for stock A along with the variance
of return on the market portfolio from Problem 1 we have:
Similarly:
V b2 46.62 ; V c2 265.0
A.2
From Problem 1 we have:
B.
B.1
According to the Sharpe single-index model, the covariance between the returns
on a pair of assets is:
SIMV ij E i E jV m
2
Using the betas for stocks A and B along with the variance of the market portfolio
from Problem 1 we have:
Similarly:
R
12
it Ri Rjt Rj
V ij t 1
12
Applying the above formula to the monthly data given in Problem 1 for securities
A, B and C gives:
C.
C.1
Using the earlier results from the Sharpe single-index model, the mean monthly
return and standard deviation of an equally weighted portfolio of stocks A, B and
C are:
1 1 1
RP u 2.946% u 6.032% u 3.556% 4.18%
3 3 3
1
2
1
2
1
2
1 2 1
2
1
2
VP
u 51.15 u 46.62 u 265.0 2 u u 25.25 u 57.43 u 49.57
3 3 3 3 3 3
8.348%
C.2
Using the earlier results from the historical data, the mean monthly return and
standard deviation of an equally weighted portfolio of stocks A, B and C are:
1 1 1
RP u 2.946% u 6.031% u 3.554% 4.18%
3 3 3
1
2
1
2
1
2
1 2 1
2
1
2
VP u 51.15 u 46.62 u 265.0 2 u u 18.46 u 61.62 u 54.08
3 3 3
3 3 3
8.374%
The answers to parts B.1 and B.2 differ for sample covariance because the Sharpe
single-index model assumes the covariance between the residual returns of
securities i and j is 0 (cov(Hi Hj ) = 0), and so the single-index form of sample
covariance of total returns is calculated by setting the sample covariance of the
sample residuals equal to 0. The sample-statistics form of sample covariance of
total returns incorporates the actual sample covariance of the sample residuals.
The answers in parts C.1 and C.2 for mean returns on an equally weighted portfolio
of stocks A, B and C are identical because the Sharpe single-index model formula
for the mean return on an individual stock yields a result identical to that of the
sample-statistics formula for the mean return on the stock.
The answers in parts C.1 and C.2 for standard deviations of return on an equally
weighted portfolio of stocks A, B and C are different because the Sharpe single-
index model formula for the sample covariance of returns on a pair of stocks yields
a result different from that of the sample-statistics formula for the sample
covariance of returns on a pair of stocks.
Chapter 7: Problem 3
Recall from the text that the Vasicek techniques forecast of security is beta ( E i 2 )
is:
V E2i1 V E21
E i2 u E1 u E i1
V E21 V E2i1 V E21 V E2i1
where E 1 is the average beta across all sample securities in the historical period (in
this problem referred to as the market beta), E i1 is the beta of security i in the
historical period, V E21 is the variance of all the sample securities betas in the
historical period and V E2i1 is the square of the standard error of the estimate of beta
for security i in the historical period.
If the standard errors of the estimates of all the betas of the sample securities in the
historical period are the same, then, for each security i, we have:
V E2i1 a
where a is a constant across all the sample securities.
a V E21
E i2 u E1 u E i1 X E 1 1 X E i1
V E21 a V E21 a
This shows that, under the assumption that the standard errors of all historical betas
are the same, the forecasted beta for any security using the Vasicek technique is
a simple weighted average (proportional weighting) of E 1 (the market beta)
and E i1 (the securitys historical beta), where the weights are the same for each
security.
Chapter 7: Problem 4
Letting the historical period of the year of monthly returns given in Problem 1 equal
1 (t = 1), then the forecast period equals 2 and the Blume forecast equation is:
Using the earlier answer to Problem 1 for the estimate of beta from the historical
period for stock A along with the above equation we obtain the stocks
forecasted beta:
Similarly:
E B2 1.023 ; E C2 1.803
Chapter 7: Problem 5
A.
The single-index model's formula for security i's mean return is
Ri = D i + E i Rm
R A = D A + E A Rm
= 2 + 1.5 x 8
= 2 + 12
= 14%
B.
The single-index model's formula for security i's own variance is:
2 2
V i = E i V m + V ei .
2 2
V A = E A V m + V eA
2 2 2 2
= 1.5 u 5 3
2 2 2
= 65.25
Similarly:
C.
The single-index model's formula for the covariance of security i with security j is
V ij = V ji = E i E j V m
2
V AB = E A E B V m
2
= 1.5 u 1.3 u 25
= 48.75
Similarly:
Chapter 7: Problem 6
A.
Recall that the formula for a portfolio's beta is:
N
E P = Xi E i
i =1
The weight for each asset (Xi) in an equally weighted portfolio is simply 1/N, where
N is the number of assets in the portfolio.
D P = Xi D i
i =1
N
V P2 E P2V m
2
Xi2V e2i
i 1
2 2 2 2
V 2
P 1.25 5 1 32 1 12 1 22 1 42
2 2
4 4 4 4
1.252 u 25 1 9 1 4 16
16
33.52
RP D P E P Rm
2.5 1.125 u 8
11.5%
Chapter 7: Problem 7
Using E i 2 0.343 0.677E i1 and the historical betas given in Problem 5 we can
forecast, e.g., the beta for security A:
E A2 0.343 0.677E A1
0.343 0.677 u 1.5
0.343 1.0155
1.3585
Similarly:
Chapter 7: Problem 8
Using the historical betas given in Problem 5 and Vasiceks formula, we can
forecast, e.g., the beta of security A:
V E2A1 V E21
E A2 u E1 2 u E A1
V E21 V E2A1 V E 1 V E2A1
0.212 u 1
0.252 u 1.5
0.252 0.212 0.252 0.212
0.0441 0.0625
u 1 u 1.5
0.0625 0.0441 0.0625 0.0441
0.4137 u 1 0.5863 u 1.5
0.4137 0.8795
1.2932
Similarly:
Given the assumptions of the Sharpe single-index model, the single-index models
expression for the covariance between the returns on a pair of securites is:
SIMV ij E i E jV m
2
V im V jm
u 2 u Vm 2
Vm Vm
2
U imV iV m U jmV jV m
u u Vm
2
Vm
2
Vm2
U imU jmV iV j
If the assumptions of both the constant correlation and single-index model hold,
then we have CCV ij SIMV ij :
U *V iV j U imU jmV iV j or U * U im U jm
This must hold for all pairs of securities, including i and j, i and k and j and k. So we
have:
U* U im U jm
U *
U imU km
U* U jmU km
The only solution to the above set of equations is:
U im U jm U km U*
Therefore, for any security i we have:
V im U imV i V m U imV i U*
Ei uV i
Vm 2
Vm
2
Vm Vm
In other words, given that all pairs of securities have the same correlation
coefficient and that the Sharpe single-index model holds, each securitys beta is
proportional to its standard deviation, where the proportion is a constant across all
U*
securities equal to .
Vm
I 2* J 0 J 1 u I1 d t or I2 dt I2* J 0 J 1 u I1
which gives:
I2* J 0 J 1 u I1 I2
Substituting the above expression into equation (1) and rearranging we get:
Ri a *
i
bi*2 u J 0 bi*1 bi*2 u J 1 u I1 bi*2 u I2 bi*3 u I3* ci
c
The first term in the above equation is a constant, which we can define as a1 . The
coefficient in the second term of the above equation is also a constant, which we
c
can define as bi1 . We can then rewrite the above equation as:
c c
Ri ai bi1 u I1 bi*2 u I2 bi*3 u I3* ci (2)
I3* T0 T1 u I1 T 2 u I2 et or I 3 et I 3* T 0 T 1 u I1 T 2 u I 2
which gives:
I 3* T 0 T 1 u I1 T 2 u I 2 I 3
Substituting the above expression into equation (2) and rearranging we get:
Ri
i i3 0
i1 i3 1
1 i2 i3 2
a c b u T b c b u T u I b* b* u T u I b* u I c
2 i3 3 i
In the above equation, the first term and all the coefficients of the new orthogonal
indices are constants, so we can rewrite the equation as:
Ri ai bi1 u I1 bi 2 u I2 bi 3 u I3 ci
Recall from the earlier chapter on the single-index model that an expression for
the covariance of returns on two securities i and j is:
V ij
2
>
E i E j E Rm Rm E j E ei Rm Rm E i E ej Rm Rm E>ei ej @ @ > @
The first term contains the variance of the market portfolio, the second two terms
contain the covariance of the market portfolio with the residuals and the last term
is the covariance of the residuals.
Given that one of the models assumptions is that the covariance of the market
portfolio with the residuals is zero and that, from the problem, the covariance of
the residuals equals a constant K, the derived covariance between the two
securities is:
V ij E i E jV m
2
K
N N N
V P2
i 1
Xi2V i2 X X V
j 1 k 1
j k jk
kz j
Recalling that the single-index models expression for the variance of a security is
V i2 E i2V m
2
V ei2 and substituting that expression and the derived expression for
covariance into the above equation and rearranging gives:
N N N N N N
V P2
i 1
Xi2 E i2V m
2
i 1
Xi2V ei2 j 1 k 1
X j Xk E j E kV m
2
X X K
j 1 k 1
j k
kz j kz j
N N N N N
X X E EV
i 1 j 1
i j i j
2
m X V
i 1
2
i
2
ei X X K
j 1 k 1
j k
kz j
N
N
2 N N N
Xi E i Xi E i V m Xi2V ei2 X X K j k
i 1 i 1 i 1 j 1 k 1
kz j
N N N
E V
2
P
2
m Xi2 V 2
ei K X j Xk
i 1 j 1 k 1
kz j
Ri ai bi1 u I1 bi 2 u I2 bi 3 u I3 ci
Since the residual ci always has a mean of zero (by construction if necessary), we
have the following expression for expected return:
Ri a i b i1 u I1 b i 2 u I 2 b i 3 u I 3
V i2
E ai bi1 u I1 bi 2 u I2 bi 3 u I3 ci ai bi1 u I1 bi 2 u I2 bi 3 u I3
2
E b I I b I
I2 bi 3 I3 I3 ci
2
i1 1 1 i2 2
Carrying out the squaring, noting that the indices are all orthogonal with each
other and making the usual assumption that the residual is uncorrelated with any
index gives us:
V i2 bi21V I21 bi22V I22 bi23V I23 V ci2
V i2 E
ai bi1 u I1 bi 2 u I2 bi 3 u I3 ci ai bi1 u I1 bi 2 u I2 bi 3 u I3
u a j b j1 u I1 b j 2 u I2 b j 3 u I3 c j a j b j1 u I1 b j 2 u I2 b j 3 u I3
>
E bi1 I1 I1 bi 2 I2 I2 bi 3 I3 I3 ci u b j1 I1 I1 b j 2 I2 I2 b j 3 I3 I3 c j @
Carrying out the multiplication, noting that the indices are all orthogonal with
each other, making the usual assumption that the residuals are uncorrelated with
any index and assuming that the residuals are uncorrelated with each other gives
us:
V ij bi1b j1V I21 bi 2 b j 2V I22 bi 3 b j 3V I23
The formula for a security's expected return using a general two-index model is:
Ri ai bi1 u I1 bi 2 u I 2
Using the above formula and data given in the problem, the expected return for,
e.g., security A is:
RA a A b A1 u I1 b A2 u I 2
2 0.8 u 8 0.9 u 4
12%
Similarly:
RB 17% ; RC 12.6%
Using the above formula, the variance for, e.g., security A is:
V A2 b A2 1V I21 b A2 2V I22 V cA
2
C. The two-index model's formula for the covariance of security i with security j is:
Using the above formula, the covariance of, e.g., security A with security B is:
For an industry-index model, the text gives two formulas for the covariance
between securities i and k. If firms i and k are both in industry j, the covariance
between their securities' returns is given by:
V ik bim bkmV m
2
bij bkj V Ij2
Otherwise, if the firms are in different industries, the covariance of their securities'
returns is given by:
V ik bim bkmV m
2
V AB b Am bBmV m
2
b A2 bB 2V I22
0.81.122 0.91.32.52
3.52 7.3125 10.8325
The second formula should be used for the other pairs of firms:
V AC b Am bCmV m
2
V BC bBm bCmV m
2
Chapter 8: Problem 7
The answers for this problem are found in the same way as the answers for problem 6,
except that now only firms B and C are in the same industry. So for firms B and C, the
covariance between their securities' returns is:
V BC bBm bCmV m
2
bB2 bC2V I22
1.10.922 1.31.12.52
3.96 8.9375 12.8975
V AB b Am bBmV m
2
V AC b Am bCmV m
2
Chapter 8: Problem 8
To answer this problem, use the procedure described in Appendix A of the text.
First, I1 is defined as being equal to I*1 , then I*2 is regressed on I1 to obtain the given
regression equation. Since dt is uncorrelated with I1 by the techniques of regression
analysis, dt is an orthogonal index to I1. So, define I2 = dt. Then express the given
regression equation as:
Now, substitute the above equation for I*2 into the given multi-index model and
simplify:
The two-index model has now been transformed into one with orthogonal indices
I1 and I2, where I1 = I*1, and I2 = dt = I*2 - 1 - 1.3 I1.
Chapter 9: Problem 1
In the table below, given that the riskless rate equals 5%, the securities are ranked
in descending order by their excess return over beta.
Ri RF R R E E i2 i
R j RF E j i E j2
Security Rank i Ri RF
Ei
i
V ei2
F i
V ei2
1 V ej2
V 2
j 1 ej
Ci
j
1 1 10 10.0000 0.3333 0.0333 0.3333 0.0333 2.5000
6 2 9 6.0000 1.3500 0.2250 1.6833 0.2583 4.6980
2 3 7 4.6667 0.5250 0.1125 2.2083 0.3708 4.6910
5 4 4 4.0000 0.2000 0.0500 2.4083 0.4208 4.6242
4 5 3 3.7500 0.2400 0.0640 2.6483 0.4848 4.5286
3 6 6 3.0000 0.3000 0.1000 2.9483 0.5848 4.3053
The numbers in the column above labeled Ci were obtained by recalling from the
text that, if the Sharpe single-index model holds:
2
i
R j RF E j
Vm
V 2
Ci j 1 ej
i E j
2
2
1 V m
j 1 V 2
ej
Thus, given that V m
2
= 10:
10 u 0.3333 3.333
C1 2.500
1 10 u 0.0333 1.333
10 u 1.6833 16.833
C2 4.698
1 10 u 0.2583 3.583
etc.
Ri RF
With no short sales, we only include those securities for which ! C i . Thus,
Ei
only securities 1 and 6 (the highest and second highest ranked securities in the
above table) are in the optimal (tangent) portfolio. We could have stopped our
R RF
calculations after the first time we found a ranked security for which i Ci ,
Ei
(in this case the third highest ranked security, security 2), but we did not so that we
R RF
could demonstrate that i C i for all of the remaining lower ranked securities
Ei
as well.
Elton, Gruber, Brown and Goetzmann 67
Modern Portfolio Theory and Investment Analysis, 6th Edition
Solutions To Text Problems: Chapter 9
Since security 6 (the second highest ranked security, where i = 2) is the last ranked
R RF
security in descending order for which i ! C i , we set C* = C2 = 4.698 and
Ei
solve for the optimum portfolios weights using the following formulas:
Ei Ri RF
Zi C*
V 2 E
ei i
Zi
Xi 2
Z
i 1
i
1
Z1 10 4.698 0.1767
30
1.5
Z2 6 4.698 0.1953
10
0.1767
X1 0.475
0.3720
0.1953
X2 0.525
0.3720
Since i = 1 for security 1 and i = 2 for security 6, the optimum (tangent) portfolio
when short sales are not allowed consists of 47.5% invested in security 1 and 52.5%
invested in security 6.
Chapter 9: Problem 2
This problem uses the same input data as Problem 1. When short sales are allowed,
all securities are included and C* is equal to the value of Ci for the lowest ranked
security. Referring back to the table given in the answer to Problem 1, we see that
the lowest ranked security is security 3, where i = 6. Therefore, we have C* = C6 =
4.3053.
Ei Ri RF
Zi C*
V 2 E
ei i
and
Zi
Xi 6
(for the standard definition of short sales)
Z
i 1
i
or
Zi
Xi 6
(for the Lintner definition of short sales)
Z
i 1
i
So we have:
1
Z1 10 4.3053 0.1898
30
1.5
Z2 6 4.3053 0.2542
10
1.5
Z3 4.667 4.3053 0.0271
20
1
Z4 4 4.3053 0.0153
20
0.8
Z5 3.75 4.3053 0.0444
10
2.0
Z6 3 4.3053 0.0653
40
Z
i 1
i 0.1898 0.2542 0.0271 0.0153 0.0444 0.0653 0.3461
Z
i 1
i 0.1898 0.2542 0.0271 0.0153 0.0444 0.0653 0.5961
0.1898 0.1898
Security 1 (i = 1) X1 0.5484 X1 0.3184
0.3461 0.5961
0.2542 0.2542
Security 6 (i = 2) X2 0.7345 X2 0.4264
0.3461 0.5961
0.0271 0.0271
Security 2 (i = 3) X3 0.0783 X3 0.0455
0.3461 0.5961
0.0153 0.0153
Security 5 (i = 4) X4 0.0442 X4 0.0257
0.3461 0.5961
0.0444 0.0444
Security 4 (i = 5) X5 0.1283 X5 0.0745
0.3461 0.5961
0.0653 0.0653
Security 3 (i = 6) X6 0.1887 X6 0.1095
0.3461 0.5961
Chapter 9: Problem 3
With short sales allowed but no riskless lending or borrowing, the optimum portfolio
depends on the investors utility function and will be found at a point along the
upper half of the minimum-variance frontier of risky assets, which is the efficient
frontier when riskless lending and borrowing do not exist. As is described in the text,
the entire efficient frontier of risky assets can be delineated with various
combinations of any two efficient portfolios on the frontier. One such efficient
portfolio was found in Problem 2. By simply solving Problem 2 using a different
value for RF , another portfolio on the efficient frontier can be found and then the
entire efficient frontier can be traced using combinations of those two efficient
portfolios.
In the table below, given that the riskless rate equals 5%, the securities are ranked
in descending order by their excess return over standard deviation.
Ri RF i
R j RF U
Security Rank i Ri R F
Vi Vj
1
1 U iU
Ci
j
1 1 10 1.00 1.00 0.5000 0.5000
2 2 15 1.00 2.00 0.3333 0.6667
5 3 5 1.00 3.00 0.2500 0.7500
6 4 9 0.90 3.90 0.2000 0.7800
4 5 7 0.70 4.60 0.1667 0.7668
3 6 13 0.65 5.25 0.1429 0.7502
7 7 11 0.55 5.80 0.1250 0.7250
The numbers in the column above labeled Ci were obtained by recalling from the
text that, if the constant-correlation model holds:
U i
R j RF
Ci u
1 U iU
1 V j
j
Ri RF
With no short sales, we only include those securities for which ! C i . Thus,
Vi
only securities 1, 2, 5 and 6 (the four highest ranked securities in the above table)
are in the optimal (tangent) portfolio. We could have stopped our calculations
R RF
after the first time we found a ranked security for which i C i , (in this case
Vi
the fifth highest ranked security, security 4), but we did not so that we could
R RF
demonstrate that i C i for all of the remaining lower ranked securities as
Vi
well.
1 Ri RF
Zi 1 U V C*
i V i
Zi
Xi 4
Z
i 1
i
1
Z2 1 0.78 0.0293
0.515
1
Z3 1 0.78 0.0880
0.55
1
Z4 0.9 0.78 0.0240
0.510
0.0440
X1 0.2375
0.1853
0.0293
X2 0.1581
0.1853
0.0880
X3 0.4749
0.1853
0.0240
X4 0.1295
0.1853
Since i = 1 for security 1, i = 2 for security 2, i = 3 for security 5 and i = 4 for security 6,
the optimum (tangent) portfolio when short sales are not allowed consists of
23.75% invested in security 1, 15.81% % invested in security 2, 47.49% % invested in
security 5 and 12.95% invested in security 6.
Elton, Gruber, Brown and Goetzmann 72
Modern Portfolio Theory and Investment Analysis, 6th Edition
Solutions To Text Problems: Chapter 9
Chapter 9: Problem 5
This problem uses the same input data as Problem 4. When short sales are allowed,
all securities are included and C* is equal to the value of Ci for the lowest ranked
security. Referring back to the table given in the answer to Problem 4, we see that
the lowest ranked security is security 7, where i = 7. Therefore, we have C* = C7 =
0.725.
To solve for the optimum portfolios weights, we use the following formulas:
1 Ri RF
Zi C*
1 U V i
V i
and
Zi
Xi 7
(for the standard definition of short sales)
Z
i 1
i
or
Zi
Xi 7
(for the Lintner definition of short sales)
Z
i 1
i
So we have:
1
Z1 1 0.725 0.0550
0.510
1
Z2 1 0.725 0.0367
0.515
1
Z3 1 0.725 0.1100
0.55
1
Z4 0.9 0.725 0.0350
0.510
1
Z5 0.7 0.725 0.0050
0. 5 10
1
Z6 0.65 0.725 0.0075
0.520
1
Z7 0.55 0.725 0.0175
0.520
7
Z
i 1
i 0.0550 0.0367 0.1100 0.0350 0.0050 0.0075 0.0175 0.2067
Z
i 1
i 0.0550 0.0367 0.1100 0.0350 0.0050 0.0075 0.0175 0.2667
0.0550 0.0550
Security 1 (i = 1) X1 0.2661 X1 0.2062
0.2067 0.2667
0.0367 0.0367
Security 2 (i = 2) X2 0.1776 X2 0.1376
0.2067 0.2667
0.1100 0.1100
Security 5 (i = 3) X3 0.5322 X3 0.4124
0.2067 0.2667
0.0350 0.0350
Security 6 (i = 4) X4 0.1703 X4 0.1312
0.2067 0.2667
0.0050 0.0050
Security 4 (i = 5) X5 0.0242 X5 0.0187
0.2067 0.2667
0.0075 0.0075
Security 3 (i = 6) X6 0.0363 X6 0.0281
0.2067 0.2667
0.0175 0.0175
Security 7 (i = 7) X7 0.0847 X7 0.0656
0.2067 0.2667
Chapter 9: Problem 6
With short sales allowed but no riskless lending or borrowing, the optimum portfolio
depends on the investors utility function and will be found at a point along the
upper half of the minimum-variance frontier of risky assets, which is the efficient
frontier when riskless lending and borrowing do not exist. As is described in the text,
the entire efficient frontier of risky assets can be delineated with various
combinations of any two efficient portfolios on the frontier. One such efficient
portfolio was found in Problem 5. By simply solving Problem 5 using a different
value for RF , another portfolio on the efficient frontier can be found and then the
entire efficient frontier can be traced using combinations of those two efficient
portfolios.
Expected utility of investment A = 1/3 u 7.5 + 1/3 u 12.5 + 1/3 u 31.5 = 17.0
Expected utility of investment B = 1/4 u 4.0 + 1/2 u 17.5 + 1/4 u 40.0 = 19.75
Expected utility of investment C = 1/5 u 0.5 + 3/5 u 31.5 + 1/5 u 144.0 = 47.8
Expected utility of investment A = 1/3 u 0.45 + 1/3 u 0.41 + 1/3 u 0.33 = 0.40
Expected utility of investment B = 1/4 u 0.50 + 1/2 u 0.38 + 1/4 u 0.32 = 0.39
Therefore, the first outcomes probability of 0.5 would have to be reduced by 0.11
to 0.39, and the second outcomes probability of 0.25 would have to be increased
by 0.11 to 0.36.
1 3 5 / 2
Given UW W 1/ 2 , then UcW W 3 / 2 and UccW W . Therefore:
2 4
3 5 / 2
W
3 1
AW 4 W
1 2
W 3 / 2
2
3
AcW W 2 0
2
3 3 / 2
W
3
RW 4
1 2
W 3 / 2
2
RcW 0
Therefore the utility function exhibits decreasing absolute risk aversion and
constant relative risk aversion.
Regarding the utility functions properties of absolute and relative risk aversion, we
have:
bc 2 e cW
AW c
bce cW
AcW 0
bc 2 WecW
RW cW
bce cW
RcW c ! 0
Therefore the utility function exhibits constant absolute risk aversion and increasing
relative risk aversion (since c < 0).
1 3 / 2
UcW W ! 0 (individual prefers more wealth to less)
2
3
UccW W 5 / 2 0 (individual is risk averse)
4
3 5 / 2
W
3 1
AW 4 W
1 3 / 2 2
W
2
3
AcW W 2 0 (decreasing absolute risk aversion)
2
e W
AW 1
e W
We W
RW W
e W
The investor will prefer the investment that maximizes expected utility of terminal
wealth. Recall that the formula for expected utility of wealth (E[U(W)]) is:
E>UW @ UW u PW
W
where each P(W) is the probability associated with each particular outcome of
wealth (W). Since UW W 0.05W 2 , we have:
Investment A:
To solve this problem, set the expected utility of investment A in Problem 10 equal
to 4.635 (the expected utility of investment B) and solve for the value of the first
outcome in investment A:
X 2 20 X 86 0
The equation above is a quadratic equation with two roots. Using the quadratic
formula, the roots are found to be 6.26 and 13.74. So, the minimum amount that
the first outcome of investment A would have to change by for the investor to be
indifferent between investments A and B would be $6.26 $5 = $1.26 (an increase),
since both investments would then provide the same level of expected utility.
Roys safety-first criterion is to minimize Prob(RP < RL). If RL = 5%, then for the
investments in Problem 1 we have:
Prob(RA < 5%) = 0.2
Thus, using Roys safety-first criterion, investments B and C are preferred over
investment A, and the investor would be indifferent to choosing either investment
B or C.
3.99% for A
5.99% for B
5.99% for C
Thus, B and C are preferred to A, but are indistinguishable from each other.
Elton, Gruber, Brown and Goetzmann 81
Modern Portfolio Theory and Investment Analysis, 6th Edition
Solutions To Text Problems: Chapter 11
Chapter 11: Problem 4
Employing Telser's criterion, we see that Project A in Problem 1 does not satisfy the
constraint Prob(Rp d 5%) d 10%. Thus it is eliminated. Between B and C, Project C
has higher expected return (7.1% compared to 7%). Thus it is preferred using
Telsers criterion of maximizing Rp subject to the constraint.
cumulative probability
outcome A B C
Roy's criterion is to minimize Prob(RP < RL). When RL = 3%, Prob(RA < 3%) = 0,
Prob(RB < 3%) = 0, and Prob(RC < 3%) = 0. So, investments A, B, and C are
indistinguishable using Roy's criterion with RL = 3%.
RN RF RUS RF
! u U N,US
VN V US
As is explained in the text, if the above inequality holds, then the foreign
investment will be attractive to a U.S. investor. RUS and RN for the foreign
countries are given in the problem's table. From the tables in the text, we have:
VN UN,US
Also, from the text tables, VUS = 13.59. Given that RF = 6%, we have:
RN RF RUS RF
u U N,US
VN V US
For Austria and France, the above inequality holds, so a U.S. investor should
consider those foreign markets as attractive investments; for Japan and the U.K.,
the above inequality does not hold, so a U.S. investor should not consider those
foreign markets as attractive investments.
To answer this question, use the formula introduced in Chapter 5 for finding the
minimum-risk portfolio of two assets:
V 22 V 1V 2 U12
X1GMV
V 12 V 22 2V 1V 2 U12
For equities, VUS = 13.59, VN = 16.70 and UN,US = 0.423. So the minimum-risk portfolio is:
GMV
X US
16.72 13.5916.70.423
13.592 16.72 213.5916.70.423
0.6734 67.34%
X NGMV 1 X US
GMV
0.3266 32.66%
For bonds, VUS = 7.90, VN = 9.45 and UN,US = 0.527. So the minimum-risk portfolio is:
GMV
X US
9.452 7.99.450.527
7.92 9.452 27.99.450.527
0.6841 68.41%
X NGMV 1 X US
GMV
0.3159 31.59%
For T-bills, VUS = 0.35, VN = 6.77 and UN,US = 0.220. So the minimum-risk portfolio is:
GMV
X US
6.772 0.356.77 0.22
0.352 6.772 20.356.77 0.22
0.9863 98.63%
X NGMV 1 X US
GMV
0.0137 1.37%
In the text, the return due to exchange-rate changes (RX) is shown to be equal to
fxt/fxt-1 - 1, where fxt is the foreign exchange rate at time t expressed in terms of the
investor's home currency per unit of foreign currency. Let fxt be the exchange rate
expressed in terms of dollars and fx*t be the exchange rate expressed in terms of
pounds. These two rates are simply reciprocals, i.e., fx*t = 1/fxt. So from the table in
the problem we have:
(1 + RX) (1 + R*X)
Period (for US investor) (for UK investor)
1 2.5/3 = 0.833 3/2.5 = 1.200
2 2.5/2.5 = 1.000 2.5/2.5 = 1.000
3 2/2.5 = 0.800 2.5/2 = 1.250
4 1.5/2 = 0.750 2/1.5 = 1.333
5 2.5/1.5 = 1.667 1.5/2.5 = 0.600
The total return to a U.S. investor from a U.K. investment is (1 + RX)(1 + RUK) 1; the
total return to a U.K. investor from a U.S. investment is(1 + R*X)(1 + RUS) 1. So:
From U.S.
Period Investment From U.K. Investment
From U.K.
Period Investment From U.S. Investment
1 5% (1.2)(1.1) 1 = 32.0%
2 5% (1)(1.15) 1 = 15.0%
3 15% (1.25)(0.95) 1 = 18.75%
4 8% (1.333)(1.12) 1 = 49.3%
5 10% (0.6)(1.06) 1 = 36.4%
V US
10 7.62 15 7.62 5 7.62 12 7.62 6 7.62
5
6.95%
V UK
12.5 7.762 5 7.762 8 7.762 19 7.762 83.3 7.762
5
38.06%
V UK
5 6.62 5 6.62 15 6.62 8 6.62 10 6.62
5
6.65%
V US
32 15.732 15 15.732 18.75 15.732 49.3 15.732 36.4 15.732
5
28.70%
This problem is essentially the same as Problem 3, except that the exchange rate is
given in indirect (yen/$) terms rather than direct ($/yen) terms. From the table in
the problem we have:
(1 + RX) (1 + R*X)
Period (for US investor) (for Japanese investor)
1 200/180 = 1.111 180/200 = 0.900
2 180/190 = 0.947 190/180 = 1.056
3 190/150 = 1.267 150/190 = 0.789
4 150/170 = 0.882 170/150 = 1.133
5 170/180 = 0.944 180/170 = 1.059
From U.S.
Period Investment From Japan Investment
From Japan
Period Investment From U.S. Investment
The answers to this problem are found in the same way as those to Problem 4.
Use the formula for the sample correlation coefficient U with five observations:
R
5
USt RUS R Jt R J
U t 1
R R
5 5
2 2
USt RUS Jt RJ
t 1 t 1
Ri
RF Rm RF E i
Thus, from the data in the problem we have:
6
RF Rm RF u 0.5 for asset 1
12
RF Rm RF u 1.5 for asset 2
3 + (9 3) u 2 = 15%
Given the security market line in this problem, for the two stocks to be fairly priced
their expected returns must be:
If the expected return on either stock is higher than its return given above, the
stock is a good buy.
Given the security market line in this problem, the two funds expected returns
would be:
Given the security market line in this problem, the riskless rate equals 0.04 (4%), the
intercept of the line, and the excess return of the market above the riskless rate
(also called the market risk premium) equals 0.10 (10%), the slope of the line.
(The return on the market portfolio must therefore be 0.04 + 0.10 = 0.14, or 14%.)
The price form of the CAPMs security market line equation is:
cov Yi Ym
Pi
1
rF
Yi Ym rF u Pm u
var Ym
Ym Pm
where rF 1 RF and Rm .
Pm
Ym Pm
From Problem 4, we have RF 0.04 and Rm 0.14 . Therefore 0.14
Pm
which gives 1.14Pm Ym .
Substituting these vales into the above security market line equation, we have:
1 cov Yi Ym
Pi Yi 1.14 u Pm 1.04 u Pm u
1.04 var Ym
1 cov Yi Ym
Yi 0.10 u Pm u
1.04 var Ym
To be rigorous, one should use the four Kuhn-Tucker conditions shown in Appendix
E of Chapter 6. To find the optimum portfolio when short sales are not allowed, we
have, for each asset i, the following Kuhn-Tucker conditions:
dT
Ui 0 (1)
dX i
X i Ui 0 (2)
Xi t 0 (3)
Ui t 0 (4)
We have already seen that, given the assumptions of the standard CAPM, setting
dT
0 gives the equilibrium first order condition for asset i, which is the standard
dX i
CAPMs security market line:
Ri
R F R m RF E i
or equivalently
Ri RF Rm RF E i 0
When short sales are not allowed, Kuhn-Tucker condition (1) implies that:
Ri RF Rm RF E i Ui 0
But, since all assets are held long in the market portfolio, Xi > 0 for each asset and
therefore, given Kuhn-Tucker condition (2), Ui = 0 for each asset. Thus, the standard
CAPM holds even if short sales are not allowed.
Using the two assets in Problem 1, a portfolio with a beta of 1.2 can be
constructed as follows:
X1 = 0.3; X2 = 0.7
Asset 3 has a higher expected return than the portfolio of assets 1 and 2, even
though asset 1 and the portfolio have the same beta. Thus, buying asset 3 and
financing it by shorting the portfolio would produce a positive (arbitrage) return of
15% 10.2% = 4.8% with zero net investment and zero beta risk.
Ri
R F R m RF E i
Substituting the given values for assets 1 and 2 gives two equations with two
unknowns:
9.4
RF R m RF u 0.8
13.4
RF R m RF u 1.3
Solving simultaneously gives:
RF 3% ; Rm 11%
Given the zero-beta security market line in this problem, the return on the zero-
beta portfolio equals 0.04 (4%), the intercept of the line, and the excess return of
the market above the zero-beta portfolios return (also called the market risk
premium) equals 0.10 (10%), the slope of the line. The return on the market
portfolio must therefore be 0.04 + 0.10 = 0.14, or 14%.
RZ has the same role in the zero-beta model as RF does in the standard model. So,
referring back to the answer to Problem 5 in Chapter 13, simply replace RF with RZ
to obtain:
cov Yi Ym
Pi
1
rZ
Yi Ym rZ u Pm u
var Ym
where rZ 1 R .
Z
As is shown in the text, the post-tax form of the CAPMs equilibrium pricing
equation is:
Ri
RF Rm RF W u G m RF u E i W u G i RF
Ri
RF 1 W Rm RF W u G m RF u E i WG i
Comparing the above general equation to the specific one given in the problem,
0.05
we see that RF 1 W 0.05 , or RF , and that W 0.24 . Therefore:
1 W
0.05
RF 0.0658 6.58%
1 0.24
Since we are given R Z and only one RF , and since R Z > RF , this situation is where
there is riskless lending at RF and no riskless borrowing. The efficient frontier will
therefore be a ray in expected return-standard deviation space tangent to the
minimum-variance curve of risky assets and intersecting the expected return axis
at the riskless rate of 3% plus that part of the minimum-variance curve of risky
assets to the right of the tangency point. This is depicted in the graph below,
where the efficient frontier extends along the ray from RF to the tangent portfolio L,
then to the right of L along the curve through the market portfolio M and out
toward infinity (assuming unlimited short sales). Note that, unless all investors in the
economy choose to lend or invest solely in portfolio L, the market portfolio M will
always be on the minimum-variance curve to the right of portfolio L.
Since both M and Z are on the minimum-variance curve, the entire minimum-
variance curve of risky assets can be traced out by using combinations (portfolios)
of M and Z. Letting X be the investment weight for the market portfolio, the
expected return on any combination portfolio P of M and Z is:
RP X R m 1 X R Z (1)
VP X 2V m2 1 X V Z2
2
(2)
RP 15 X 51 X
(3)
10 X 5
Substituting the given values for V m and V Z into equation (2) gives:
VP X 2 u 22 2 1 X u 8 2
2
Using equations (3) and (4) and varying X (the fraction invested in the market
portfolio M) gives various coordinates for the minimum-variance curve; some of
them are given below:
RP 5 7 9 11 13 15 20 25
The zero-beta form of the security market line describes equilibrium beta risk and
expected return relationship for all securities and portfolios (including portfolio L)
except those combination portfolios composed of the riskless asset and tangent
portfolio L along the ray RF - L in the above graph:
Ri
RZ Rm RZ E i
5 10 E i
The equilibrium beta risk and expected return relationship for any combination
portfolio C composed of the riskless asset and tangent portfolio L along the ray
RF - L in the above graph is described by the following line:
RC RF
R L RF u E
EL
C
Ri
RF R m RF G m RF W E i G i RF W
Ri
RF R m RF E i
Assume that the post-tax model holds instead of the standard model, and G m RF .
For a stock with G i RF W ! 0 , the institution that uses the post-tax model would
correctly believe that the stock has a higher expected return than the stocks
return expected by the institution using the standard model. Similarly, for a stock
with G i RF W 0 , the institution that uses the post-tax model would correctly
believe the stock has a lower expected return than the stocks return expected by
the institution using the standard model.
If the post-tax model holds, then the institution using that model would correctly
believe that the equilibrium expected returns for the two stocks are:
The institution using the standard model would incorrectly believe that the stocks
equilibrium expected returns are:
RA 4 14 4 u 1.0
4 10 14%
RB 4 14 4 u 1.0
4 10 14%
The institution using the post-tax model would tend to buy stock A and sell stock B
short. Of course, residual risk puts a limit to the amount of unbalancing the
institution would do. But by some unbalancing, the institution earns an excess
return.
The institution using the standard model would be indifferent between the two
stocks. However, by buying stock B, the institution loses excess return.
Xi
ARB
i u1 0 (1)
a ARB X
i
ARB
i ai 0 (2)
b ARB X
i
i
ARB
bi 0 (3)
Since the above portfolio has zero net investment and zero risk with respect to the
given two-factor model, by the force of arbitrage its expected return must also be
zero:
R ARB X
i
ARB
i Ri 0 (4)
From a theorem of linear algebra, since the above orthogonality conditions (1), (2)
and (3) with respect to the X iARB result in orthogonality condition (4) with respect to
the X iARB , R i can be expressed as a linear combination of 1, ai and bi:
Ri O0 u 1 O1ai O2 bi (5)
Xi
Z
i 1
aZ X i
Z
i ai 0
bZ X i
Z
i bi 0
RZ X i
Z
i Ri O0 X iZ O1 X iZ ai O2 X iZ bi
i i i
O0
Xi
M
i 1
aM X i
M
i ai 1
bM Xi
M
i bi 0
RM X i
M
i Ri O0 X iM O1 X iMai O2 X iMbi
i i i
O0 O1
or
O1 R M O0 RM RZ
Xi
C
i 1
aC X i
C
i ai 0
bC X i
C
i bi 1
RC X i
C
i Ri O0 X iC O1 X iC ai O2 X iC bi
i i i
O0 O2
or
O2 R C O0 RC RZ
Substituting the derived values for O0, O1 and O2 into equation (5), we have:
Ri
R Z R M R Z u ai R C R Z u bi
In the graph below, the efficient frontier with riskless lending but no riskless
borrowing is the ray extending from RF to the tangent portfolio L and then along
the minimum-variance curve through the market portfolio M and out toward
infinity (assuming unlimited short sales). All investors who wish to lend will hold
tangent portfolio L in some combination with the riskless asset, since no other
portfolio offers a higher slope. Furthermore, unless all investors lend or invest solely
in portfolio L, the market portfolio M will be along the minimum-variance curve to
the right of portfolio L, since the market portfolio is a wealth-weighted average of
all the efficient risky-asset portfolios held by investors, and no rational investor
would hold a risky-asset portfolio along the curve to the left of L.
The expected return on a zero-beta asset is the intercept of a line tangent to the
market portfolio, and the zero-beta portfolio on the minimum-variance frontier
must be below the global minimum variance portfolio of risky assets by the
geometry of the graph. Furthermore, by the geometry of the graph, since the risk-
free lending rate is the intercept of the line tangent to portfolio L, and since L is to
the left of M on the minimum-variance curve, the risk-free lending rate must be
below the expected return on a zero-beta asset.
Assume the same situation as in Problem 5. The investor who believes in the
standard (pre-tax) CAPM expects a return of 14% on either security. You expect a
return before taxes of 15% on stock A and 13% on stock B. If your tax factor was
below the aggregate tax factor (W lower than 0.25) then you should buy stock B
from the other investor and sell that investor stock A. The fact that this will lead to
higher after-tax cash flows for you is straightforward.
This problem can be answered directly by using the equation developed for non-
marketable assets. The equation also holds for deleted assets, with the subscript H
now standing for those assets that were left out:
R m RF P
Ri RF u cov Ri Rm H cov Ri RH
P
H cov Rm RH
Pm
Vm
2
Pm
1.) Whether or not the returns on the aggregate of all bonds are negatively or
positively correlated with the returns on the aggregate of all stocks;
2.) The correlation between the returns on a particular stock and the returns on
the aggregate of all bonds.
From the above equation, if returns on stocks and bonds are generally positively
correlated (as empirical evidence shows), then the denominator in the second
term of the equation will tend to lower the expected return on any stock. If the
return on a particular stock is negatively correlated with bonds, that will further
lower the stocks expected return. However, if the stock is positively correlated with
bonds, this will offset the effect of positive correlation between all stocks and
bonds and may actually result in a higher expected return for the stock.
That is NOT a valid test of the theory, and the empirical evidence IS consistent with
the theory. If high-beta stocks always gave higher returns, then they would be less
risky than low-beta stocks. It is precisely because the returns on high-beta stocks
are more risky, and hence sometimes below and sometimes above the returns on
low-beta stocks, that high-beta stocks have higher expected (and over long
periods of time higher actual) returns.
Let:
R Ai = the expected percentage change in alcoholism in city i;
R G = the expected percentage change in the price of gold;
R P = the expected percentage change in professors salaries.
Then we have:
cov R Ai RP
R Ai
RG RP RG u var RP
The above equation is exactly parallel to the zero-beta CAPM equation, with
expected percentage change in alcoholism in a city playing the role of the
expected return on a security. The analogy between variables is seen from:
cov Ri Rm
Ri
RZ Rm RZ u var Rm
Therefore, tests exactly parallel to those employed in the text can be used.
Since the equality shown in equation (15.7) in the text only holds for an efficient
portfolio, if the market portfolio is inefficient the equality will not hold and instead
we have:
OV km z R k RF
The remaining proof follows the proof shown in the text below equation (15.7), but
with the not-equal sign replacing the equal sign in all the remaining equations in
the proof.
If the post-tax form of the CAPM holds, then the real relationship as a cross-
sectional regression model is:
R i RF J 0 J 1E i J 2 G i RF H i
If the standard CAPM security market line is tested, the cross-sectional regression
model is:
R i RF J 0 J 1E i H i
From the text we know that three points determine a plane. The APT equation for
a plane is:
R i O 0 O1bi1 O2 bi 2
Assuming that the three portfolios given in the problem are in equilibrium (on the
plane), then their expected returns are determined by:
12 O0 O1 0.5O2 (a)
The above set of linear equations can be solved simultaneously for the three
unknown values of O0, O1 and O2. There are many ways to solve a set of
simultaneous linear equations. One method is shown below.
Ri 10 bi1 2bi 2
According to the equilibrium APT plane derived in Problem 1, any security with
b1 = 2 and b2 = 0 should have an equilibrium expected return of 12%:
Assuming the derived equilibrium APT plane holds, since portfolio D has bD1 = 2 and
bD2 = 0 with an expected return of 10%, the portfolio is not in equilibrium and an
arbitrage opportunity exists.
1
2X A 1 or X A
2
1
X A 0.7 X B
2
1 1
Since X A , XB 0 and X C 1 X A X B .
2 2
1 1
RE X A R A X B RB X C RC u 12 0 u 13.4 u 12 12%
2 2
So now we have two portfolios with exactly the same risk: the target portfolio D
and the equilibrium replicating portfolio E. Since they have the same risk (factor
loadings), we can create an arbitrage portfolio, combining the two portfolios by
going long in one and shorting the other. This will create a self-financing (zero net
investment) portfolio with zero risk: an arbitrage portfolio.
Xi
ARB
i X EARB X DARB 1 1 0 (zero net investment)
1 1
Xi
ARB
i X AARB X BARB X CARB X DARB
2
0 1 0 (zero net investment)
2
b ARB1 X i
ARB
i bi1
b ARB 2 X i
i
ARB
bi 2
R ARB X i
i
ARB
Ri
From the text we know that three points determine a plane. The APT equation for
a plane is:
R i O 0 O1bi1 O2 bi 2
Assuming that the three portfolios given in the problem are in equilibrium (on the
plane), then their expected returns are determined by:
12 O0 O1 O2 (a)
Solving for the three unknowns in the same way as in Problem 1, we obtain the
following solution to the above set of simultaneous linear equations:
O0 8 ; O1 6 ; O2 2 ;
Ri 8 6bi1 2bi 2
According to the equilibrium APT plane derived in Problem 3, any security with
b1 = 1 and b2 = 0 should have an equilibrium expected return of 12%:
Assuming the derived equilibrium APT plane holds, since portfolio D has bD1 = 1 and
bD2 = 0 with an expected return of 15%, the portfolio is not in equilibrium and an
arbitrage opportunity exists.
bE 2 X A b A2 X B bB 2 1 X A X B bC2 X A 2 X B 31 X A X B b D2 0
1 1
XA XB
2 2
4X A 5X B 3
1 1 1
XA , XB and X C 1 X A X B .
3 3 3
1 1 1
RE X A R A X B RB XC RC u 12 u 13 u 17 14%
3 3 3
So now we have two portfolios with exactly the same risk: the target portfolio D
and the equilibrium replicating portfolio E. Since they have the same risk (factor
loadings), we can create an arbitrage portfolio, combining the two portfolios by
going long in one and shorting the other. This will create a self-financing (zero net
investment) portfolio with zero risk: an arbitrage portfolio. In equilibrium, an
arbitrage portfolio has an expected return of zero, but since portfolio D is not in
equilibrium, neither is the arbitrage portfolio containing D and E, and an arbitrage
profit may be made.
We need to short sell either portfolio D or E and go long in the other. The question is:
which portfolio do we short and which do we go long in? Since both portfolios
have the same risk and since portfolio D has a higher expected return than
portfolio E, we want to go long in D and short E; in other words, we want X DARB 1
and X EARB 1. This gives us:
X
i
ARB
i X DARB X EARB 1 1 0 (zero net investment)
1 1 1
Xi
ARB
i X AARB X BARB X CARB X DARB 1 0 (zero net investment)
3 3 3
b ARB1 X i
ARB
i bi1
b ARB 2 X i
i
ARB
bi 2
R ARB X i
i
ARB
Ri
K
Ri O0 O k bik
k 1
where O0 is the return on the riskless asset, if it exists.
The last number, 1.59, enters because the stock is a construction stock.
A.
From the text we know that, for a 2-factor APT model to be consistent with the
standard CAPM, O j
R m RF E Oj . Given that R m RF 4 and using results from
Problem 1, we have:
1 4 E O1 or E O1 0.25 ; 2 4 E O 2 or E O 2 0.5 .
B.
From the text we know that E i bi1E O1 bi 2 E O 2 . So we have:
C.
Assuming all three portfolios in Problem 1 are in equilibrium, then we can use any
one of them to find the risk-free rate. For example, using portfolio A gives:
RA
Rf R m RF E A or RF
R A R m RF E A
RF 12 4 u 0.5 10%
The simplest trading strategy would be to buy a stock at the opening price on the
day that the heard on the street column indicates analysts have reported
positive recommendations and to short sell it if analysts have reported negative
recommendations. Since any stock price effect occurs very shortly after the news
is released, the stock position could be unwound after five days. Naturally, in
examining returns from this strategy, purchases and sales would have to be
adjusted for transactions costs. The results of Davies and Canes suggest that
round-trip transactions costs must be less than two percent and perhaps less than
one percent for this rule to produce excess returns. In testing this strategy, we
would have to be sure to adjust the returns for risk. Following this strategy will lead
to a changing portfolio of stocks being held over time. Either the beta or standard
deviation of this portfolio could be used as a risk measure.
See the section in the text entitled Relative Strength for the answer to this
question.
There are several ways this rule could be tested. One way would be to rank all
stocks by their P/E ratios, select the X percent (e.g., 20%) of the stocks with the
lowest P/E ratios, then select from that group the Y percent (e.g., 20%) with the
largest five-year growth rates. After then making sure that transactions costs are
included, risk-adjusted excess returns for the final group could be obtained and
examined using one of the methodologies outlined in the text.
You could test that by following any of the test methodologies outlined in the text
for semi-strong-form efficiency, where day zero (the event day) is defined as the
day at which the block of stocks becomes available for trading.
Recall that the zero-beta CAPM leads to lower expected returns for high-beta
(above 1) stocks and higher expected returns for low-beta stocks than does the
standard CAPM. If we were testing a phenomenon that tended to occur for low-
beta stocks and not for high-beta stocks, then the zero-beta CAPM could show
inefficiency while the standard CAPM showed efficiency.
As in Problem 6, you could test that by following any of the test methodologies
outlined in the text for semi-strong-form efficiency, where day zero (the event
day) could be defined either as the day of retirement or as the day the
retirement is first announced to the public.
Since the companys growth rate of 10% extends into the future indefinitely, use
the constant-growth model to value its stock:
D1 D0 1 g 0.55 u 1.1
P0 $15.13
kg kg 0.14 0.10
D1 1
P0 $20.00
k rb 0.12 0.14 u 0.5
Solving equation (18.5b) in the text for k (the required rate of return) we have:
D1 1
k rb 0.14 u 0.5 0.103 10.3%
P0 30
Solving equation (18.5b) in the text for r (the rate of return on new investment) we
have:
D 1 1 1
r k 1 u 0.12 u 0.207 20.7%
P0 b 60 0.5
So the rate of return on new investment would have to change from 14% to 20.7%,
an increase of 6.7 percentage points.
This problem can be solved using the two-period growth model shown in the text,
where the first growth period is 5 years with a growth rate of 10% (g1) followed by a
growth rate of 6% (g2) indefinitely:
D11 g1
5 t 1
P5
P0 1 k
t 1
t
1 k 5
D6
5
D11 g1
t 1
k g2
1 k
t 1
t
1 k 5
1 g 5 D6
1 1
1 k k g2
D1
k g1 1 k 5
1 g 5 D6
1 1
1 k k g2
D0 1 g1
k g1 1 k 5
Recognizing that the dividend at the end of period 6 is equal to the dividend at
the end of period 5 compounded 1 period at g2 and then adjusted by a factor of
0.5/0.3 to reflect the increased dividend payout rate, we have:
0.5
D6 D5 1 g2 u
0.3
0.5
D0 1 g1 1 g2 u
5
0.3
0.5
0.55 u 1.1 u 1.06 u
5
0.3
$1.565
1.1 5 1.565
1
1.14 0.14 0.06
P0 0.55 u 1.1u
0.14 0.10 1.145
0.16355 19.563
0.605 u
0.04 1.925
2.474 10.163
$12.64
This problem can be solved using the three-period growth model shown in the text,
where the first growth period is 5 years with a growth rate of 10% (g1) followed
each year by linearly declining growth rates (g2, g3, g4 and g5) over a second
period of 4 years down to a 6% steady-state growth rate (gs) indefinitely thereafter.
Since the growth rate is declining linearly over the 4-year period, the annual
10 6
decline is 0.8 percentage points per year. So we have g1 = 10% (first 5
5
years), g2 = 9.2% (year 6), g3 = 8.4% (year 7), g4 = 7.6% (year 8), g5 = 6.8% (year 9)
and gS = 6% (year 10 and thereafter), and the model is:
1 g 5
1 1
9
1 k Dt P9
P0 D1
t 6 1 k 1 k 9
k g1 t
1 g 5 t 4
D
1 1
D 5
1 g j 10
D1
1 k 9 k gS
j 2
k g1 t 6 1 k t 1 k 9
1 g 5 t 4
D10
1 1
D 0 1 g 1
5
1 g j
1 k k gS
9
D0 1 g1
j 2
k g1 t 6 1 k t
1 k 9
0.5
D10 D9 1 g S u
0.3
5
1 g u 1 g u 0.3
0.5
D0 1 g1 u
5
j S
j 2
0.5
0.55 u 1.1 u 1.092 u 1.084 u 1.076 u 1.068 u 1.06 u
5
0.3
$2.129
So we have:
1 g 5 t 4
1 1
9 0 D 1 g 1 5
1 g j D10
1 k k g6
D0 1 g1
j 2
P0
k g1
t 6 1 k t
1 k 9
1.1 5
1
1.14 0.55 u 1.1 u 1.092 0.55 u 1.1 u 1.092 u 1.084
5 5
0.55 u 1.1u
0.14 0.10 1.146 1.147
0.55 u 1.1 u 1.092 u 1.084 u 1.076 0.55 u 1.1 u 1.092 u 1.084 u 1.076 u 1.068
5 5
1.148 1.149
2.129
0.14 0.06
1.149
2.474 0.441 0.419 0.396 0.371 8.184
$12.29
Solving equation (18.5b) in the text for k (the expected rate of return) we have:
D1 1
k rb 0.14 u 0.5 0.181 18.1%
P0 9
Since the companys growth rate of 10% extends into the future indefinitely, use
the equation (18.6) in the text from the constant-growth model:
D1 D0 1 g
k g g
P0 P0
0.55 u 1.1
0.1
9
0.167 16.7%
This problem can be solved using the two-period growth model shown in the text,
where the first growth period is 10 years with a growth rate of rb = g1 followed by a
growth rate of 5% (g2) indefinitely. The model is:
1 g 10 D11
1 1
P0
D1 u
1 k k g2
k g1
1 k 10
D11 D10 1 g2
D11 g1 1 g2
9
1u 1.07 u 1.05
9
$1.93
So we have:
1.07 10 1.93
1
1.12 0.12 0.05
P0 1u
0.12 0.07 1.1210
7.33 8.88
$16.21
This problem can be solved iteratively by substituting various values for k into the
first formula shown in the answer for Problem 9. By trial and error the solution is
k = 9.6%.
As with Problem 10, this problem can be solved iteratively by substituting various
values for the length of the first growth period into the first formula shown in the
answer for Problem 9. By trial and error the solution is 24 years.
The solution to this problem is a general form of the model shown in the answer to
Problem 6:
1 g N1
1 1
N1 N2
1 k Dt
P0 D0 1 g1 u PN1 N2
t N11 1 k 1 k
N1 N2
k g1 t
1 g N1 N2
g gS DN1 N2 1
N1 N2 D0 1 g1 1 g1 j u 1
N1
1 1
1 k j 1 N2 1 k gS
D0 1 g1 u
k g1
t N1 1 1 k t
1 k 1 2
N N
where
N2 = the number of years in the second growth period of linearly changing growth
rates
gS = the annual steady-state growth rate after the second period of linearly
changing growth rates
Note that the step value for linearly changing rates from g1 to gS is
(g1 gS) / (N2 + 1), not (g1 gS) / N2.
Elton, Gruber, Brown and Goetzmann 122
Modern Portfolio Theory and Investment Analysis, 6th Edition
Solutions To Text Problems: Chapter 18
Elton, Gruber, Brown and Goetzmann
Modern Portfolio Theory and Investment Analysis, 6th Edition
Solutions to Text Problems: Chapter 19
A.
If earnings follow a mean-reverting process with no trend or cycle, the following
exponential smoothing model could be used to forecast future earnings:
Et
Et 1 a Et Et 1
where
B.
If earnings follow a mean-reverting process with a trend but no cycle, either
smoothed earnings plus the trend or smoothed earnings times the trend could be
used, depending on whether the trend was additive or multiplicative. For example,
with an additive trend the forecast would be:
Et g
t
where
Et E t 1 >
t 1 a Et Et 1 g
g t 1 ; @
g t is the time-t estimate of the trend.
See footnote 7 in the text for further details on this technique.
C.
If earnings follow a mean-reverting process with a trend and a cycle, then the
forecast is smoothed earnings adjusted for the trend and the cycle. For example,
with an additive trend and a multiplicative cycle the forecast would be:
E t
t u ft
g
where ft is the time-t estimate of the cycle.
If there was a strong relationship between a firms earnings and the overall
industrys and economys earnings, then, for example, a linear model could be
estimated:
Ei a bE I cE E
where
YES. Mean reversion could be present in the industrys and economys earnings,
too.
We can use the cash flows bonds A and B to replicate the cash flows of bond C.
Let YA be the fraction of bond A purchased and YB be the fraction of bond B
purchased. (Note that these are not investment weights that sum to 1.) Then we
have:
Solving the above two equations simultaneously gives YA= YB = 1/2. So buying 1/2
of bond A and 1/2 of bond B gives the same cash flows as buying 1 bond C (or,
equivalently, buying 1 bond A and 1 bond B gives the same cash flows as buying
2 of bond C). Therefore, if the Law of One Price held, the bonds current prices
would be related as follows:
1/2 PA + 1/2 PB = PC
But, since we are given that PA = $970, PB = $936 and PC = $980, we have instead:
Given that the future cash flows of the portfolio of bonds A and B are identical in
timing and amount to those of bond C, and assuming that all three bonds are in
the same risk class, an investor should purchase 1 bond A and 1 bond B rather
than 2 of bond C.
A.
A bonds current yield is simply its annual interest payment divided by its current
price, so we have:
We can find y iteratively by trial and error, but the easiest way is to use a financial
calculator and input the following:
PV = 960
PMT = 100
FV = 1000
N=5
In general, the nominally annualized spot rate for period t (S0t) is the yield to
maturity for a t-period zero-coupon (pure discount) instrument:
F
P0 t
S0t
1
2
where P0 is the zeros current market price, F is the zeros face (par) value, and t
is the number of semi-annual periods left until the zero matures.
The zero-coupon bonds in this problem all have face values equal to $1,000.
So we have:
1000
960 1
S 01 0.0833 (8.33%)
S 01
1
2
1000
920 = 2
S02 0.0851 (8.51%)
S 02
1
2
1000
885 = 3
S03 0.0831 (8.31%)
S 03
1
2
1000
855 = 4
S04 0.0799 (7.99%)
S 04
1
2
The nominally annualized implied forward rates (ft,t+j) can be obtained from the
above spot rates. A general expression for the relationship between current spot
rates and implied forward rates is:
1
S0,t j j
t j
1
2 1 u 2
ft ,t j
t
1 S0,t
2
where t is the semi-annual period at the end of which the forward rate begins, j is
the number of semi-annual periods spanned by the forward rate, and both t and
j are integers greater than 0.
S 2
1 02
2 1.04262
f12 1 u 2 1u2 0.0870 (8.70%)
1 1.04171
1 S01
2
S03
3
1
2 1 u 2 1.04163
f23 1 u 2 0.0792 (7.92%)
2 1.04262
1 S02
2
S04
4
1
2 1 u 2 1.04004
1 u 2
f34 3
1.04163
0.0704 (7.04%)
1 S03
2
If instead we wanted the expected spot yield curve one period from now under
the pure expectations theory, we can set t equal to 1 and vary j from 1 to 3 in the
preceding equation:
S 2
1 02
2 1.04262
S12 f12 1 u 2 1 u 2 0.0870 (8.70%)
1 1.04171
1 S01
2
1
S03 2
3
1 1
2 1.0416
3 2
S13 f13 1 u 2 1
1 u 2 0.0831 (8.31%)
1.0417
1
1 S01
2
1
S04 3
4
1 1
2 1.0400
4
3
S14 f14 1 u 2
1 u 2
0.0789 (7.89%)
1.0417
1
1
1 S01
2
We can use the cash flows bonds A and B to replicate the cash flows of bond C.
Let YA be the fraction of bond A purchased and YB be the fraction of bond B
purchased. Then we have:
t = 2: $1,080 YA + $0 YB = $1,120
1,120 28 308
YA
1,080 27 297
28 3,240 2,240
120 80 u
27 27 27 1,000 27 1,000 10
YB u
1,100 29,700 27 29,700 29,700 297
27
So buying 308/297 of bond A and 10/297 of bond B gives the same cash flows as
buying 1 bond C (or, equivalently, buying 308 of bond A and 10 of bond B gives
the same cash flows as buying 297 of bond C). Therefore, if the Law of One Price
held, the bonds current prices would be related as follows:
308/297 PA + 10/297 PB = PC
But, since we are given that PA = $982, PB = $880 and PC = $1,010, we have instead:
The Law of One Price does not hold. For the Law of One Price to hold, bond C
would have to sell for $1,048.
If the Law of One Price holds, then the same discount rate (which is a spot rate)
applies for the cash flows in a particular period for all three bonds. Also, in the
presence of taxes, the price of each bond must equal the sum of the present
values of its future after-tax cash flows, where the present values are calculated
using the spot rates.
Each bonds capital gain or loss is simply its principal (par) value minus its price.
Given that each bond has a par value of $1,000, bond A has a capital gain of
$1,000 $985 = $15, bond B has a capital gain of $1,000 $900 = $100, and bond
C has a capital loss of $1,000 $1,040 = $40.
Given that the periods shown are annual, that taxes must be paid on capital gains
and can be deducted on capital losses, and that the capital gain or loss tax rate
is one-half of the ordinary income tax rate, we need to find the discount factors
and ordinary income tax rate that makes the following set of equations hold
simultaneously:
T
$80 u 1 T u d2 $80 u 1 T u d4 $15 u u d4 $1,000 u d4 $985
2
T
$100 u 1 T u d 2 100 u u d2 $1,000 u d 2 $900
2
T
$120 u 1 T u d 2 $120 u 1 T u d4 $40 u u d4 $1,000 u d4 $1,040
2
where
1
d2 2
= the two-semi-annual-period (one-year) discount factor;
S02
1
2
1
d4 4
= the four-semi-annual-period (two-year) discount factor.
S04
1
2
The duration formula shown in the text for annual payments can easily be
modified to reflect semi-annual payments as follows:
T
CFt u t
t 1 i
t
1
2
D
2 u P0
10
50 u t 1000 u 10
t 1
t
0.10 0.10
10
1 1
2 2 8.1
D 4.05 years.
2 u 1000 2
T CFt u t
1 i
1
t
D
t
P0
T 100 u t 1000 u T
1 0.10
t
1 0.10
T
t 1
D
1000
T D
10 6.76
8 5.87
5 4.17
3 2.74
Since there is just one equation with two unknowns, there are an infinite number
of solutions (portfolios) that will satisfy the equation. Either XA or XB can be
arbitrarily set and then the remaining weights solved for. Three of the infinite
number of solutions are:
Since in this problem there are three bonds with three sets of cash flows to meet
the three liabilities, we have three equations with three unknowns and therefore
one unique solution. In a more realistic situation, there would be many more
bonds than the number of liabilities (many more unknowns than the number of
equations) and thus there would be an infinite number of solutions. In that case,
the linear programming procedure shown in the texts Appendix B would be
required to find the least-cost solution.
At t = 3: $0 YA + $1,100 YB + $0 YC = $550
Assuming fractional purchases may be made, the cost of the bond portfolio is
then:
Ri Ri
Di
Dm
Rm R m ei
If the yield curve is flat at 10%, then the first periods expected return is 10% for
each of the three bonds. Since the market portfolio is a weighted average of the
three bonds, the market portfolio also has an expected return of 10%. The
duration of the market portfolio is a weighted average of the three bonds
durations. Since the three bonds are assumed to be of equal value, the value-
weighted market portfolio is also an equally weighted portfolio. Therefore, the
duration of the market portfolio is:
5
RA 10% u Rm 10% e i
9
10
RB 10% u Rm 10% e i
9
12
RC 10% u Rm 10% e i
9
We have seen in an earlier chapter that, under the assumptions of the Sharpe
single-index model, the covariance between the returns on any pair of securities
i and j is:
V ijj E i E jV m
2
Making the same assumptions as those for the Sharpe single-index model and
D
recognizing that i in the bond single-index model (equation (21.6)) is
Dm
analogous to Ei in the Sharpe single-index model, the covariance between the
returns on any pair of bonds i and j is:
Di Dj
V ijj u uV m
2
Dm Dm
Therefore we have:
DA DB 5 10
V AB u uV m
2
u uV m
2
Dm Dm 9 9
DA DC 5 12
V AC u uV m
2
u uV m
2
Dm Dm 9 9
DB DC 10 12
V BC u uV m
2
u uV m
2
Dm Dm 9 9
Although selling calls today would generate a positive cash flow for the client now,
the client would lose the potential profit he would make if the stock were to
appreciate in value, because the stock would be called away from the client.
Thus, there is a potential opportunity cost of engaging in that strategy.
The thicker line in the above diagram represents the profit from the combination. If
the options finish at the money, where the stock price at their expiration equals
their strike prices of $50, the profit would be $17 (a $17 loss). For the combination
to have a positive profit, the stock must either be below $41.50 or above $67 on
the day the options expire.
Algebraically, letting the stock price on the expiration date = P, the profit is:
P d $50 :
Profit = $17 + 2 u ($50 P) = $83 2P (since only the two put options would be
exercised)
$50 < P:
Profit = $17 + P $50 = P $67 (since only the call option would be exercised)
The profit diagram of writing the two $45 calls appears as follows:
The thicker line in the above diagram represents the profit from the combination. If
the $40 call option finishes out of or at the money, where the stock price at its
expiration is below or equal to that options strike price of $40, the profit would be
$2, because none of the options would be exercised and therefore the profit is
simply the net profit from buying the $40 call option ( $8) and selling the two $45
call options ($10). If the two $45 call options finish at the money, where the stock
price on their expiration equals their strike price of $45, the profit would be $7,
equal to the net profit of $2 from buying and writing the options plus the $5 gain
from exercising the $40 call option. $7 is the maximum profit because, at stock
prices higher than $45, although exercising the $40 call option continues to
contribute a gain, the two $45 call options that were sold will be exercised against
the seller and therefore contribute twice the loss, so the profit declines, reaching
zero at a stock price of $52. (At a stock price of $52, the profit from the $40 call will
be $12 $8 = $4 and the profit from the two $45 calls will be $14 + $10 = $4,
giving a total profit of 0.) If the stock price is greater than $52 on the expiration
date, the profit will be negative (a loss).
Algebraically, letting the stock price on the expiration date = P, the profit is:
P d $40 :
Profit = $2 (since no options would be exercised)
$40 P $45 :
Profit = $2 + P $40 (since only the $40 call option would be exercised)
$45 d P :
Profit = $2 + P $40 2 u (P $45) = $52 P (since all options would be exercised)
From the text, we know that a is the lowest number of upward moves in the stock
price at which the call takes on a positive value at expiration (finishes in the
money), u is the size of each up movement, d is the size of each down movement
and n is the number of periods remaining to the options expiration. Given that the
options exercise price (E) is $60 and the current stock price (S0) is $50, we need to
solve for the minimum integer a such that:
S0 u u a u d 1a ! E
So we have:
where
r d 1.1 0.9
P 0.67
ud 1.2 0.9
u 1.2
Pc uP u 0.67 0.73
r 1.1
So we have:
E
C S0 Nd1 Nd2
e rt
We are given:
S 1 95 1 2
ln 0 r V 2 u t ln 0.08 u 0.36 u
d1 E 2 105 2 3 0.073
0.149
V t 2 0.490
0.60 u
3
S 1 95 1 2
ln 0 r V 2 u t ln 0.08 u 0.36 u
E 2 105 2 3 0.167
d2 0.341
V t 2 0.490
0.60 u
3
Nd1 N0.149 0.560
Nd2 N 0.341 0.367
$105
C $95 u 0.560 2
u 0.367 $16.67
0.08u
3
e
The no-arbitrage condition for stock-index futures appears in the text as:
F
P PV D
1 R
Given that F = $200, P = $190, R = 6%, and PV(D) = $4, we have:
$200
$188.68 ! $190 $4 $186
1.06
so the futures are overpriced relative to the underlying index.
Therefore, the arbitrage would involve selling the futures, borrowing the present
value of the futures price and the present value of the dividends at 6% for six
months, using some of the borrowed funds to buy the index today (t = 0), and
keeping the remainder as arbitrage profit. Six months from now (t = 1), receive the
futures price for the index, receive the future value of the dividends, and use the
proceeds to pay off the loan. The cash flows are as follows:
t=0 t=1
So the arbitrage profit is $2.68 per futures contract. If the present value (at t = 0) of
transactions costs is $2.68 or greater then the arbitrage opportunity is negated.
Yes. Farmers need to be assured that they can sell their crops at harvest time,
regardless of market conditions, so that they can make planting and farm
equipment decisions in advance of the harvest. Even if both farmers and General
Mills believe that the spot price at the expiration of the futures contract will be
higher than the futures contract price (so that the farmers would get more money
selling their crops later on at the spot price than by selling futures), futures
contracts make sense economically to the farmers, since selling futures now
eliminates the uncertainty of selling their crops later.
General Mills needs to ensure a steady supply of wheat for their products
regardless of market conditions, and knowing the price of wheat in advance helps
in making pricing and working capital decisions. So a futures contract makes
economic sense from their point of view as well, even if they share the same
distributional assumptions as the farmers that spot wheat prices will be lower at the
expiration of the futures contract than the futures contract price, since buying
futures now eliminates the uncertainty of the cost and availability of wheat later.
One equation for interest rate parity that appears in the text is:
RD
F
S
1 R F 1
where RD is the domestic interest rate, RF is the foreign interest rate, F is the
domestic futures price for one unit of foreign currency, and S is the spot exchange
rate expressed as domestic currency per unit foreign currency; i.e., both F and S
are expressed in direct terms. From a U.S. viewpoint, the quotes given in the
problem are in indirect terms, so, if RD is the U.S. rate and RF is the rate for Japan,
then, from the problem, F = 1/115 and S = 1/120. So solving the above equation for
the U.S. rate gives:
1
R D 115 u 1.04 1
1
120
120
u 1.04 1
115
0.0852 8.52%
Assume you match the durations (interest rate sensitivities) of long-term and short-
term bond futures by holding them long or short in the necessary proportion.
Assuming a normal yield curve, you believe that long-term rates will fall relative to
short-term rates. If the market does not share your belief today, and if long-term
rates fall and short-term rates rise, then the prices of long-term bonds and long-
term bond futures will rise and the prices of short-term bonds and short-term bond
futures will fall. Therefore, you want to be long in long-term bond futures and short
in short-term bond futures. If instead the entire yield curve shifted up, short-term
rates would have to rise more than long-term rates for the spread to narrow, so the
above position would still be profitable. If the entire yield curve shifted down, long-
term rates would have to fall more than short-term rates for the spread to narrow,
so the above position would still be profitable.
Assuming that a futures market exists for corporate bonds, sell futures contracts to
deliver $100 million of 19-year corporates one year from today. In one year, close
out your futures position by delivering your 19-year corporate bonds; from your
viewpoint today, your 20-year corporates have thus been shortened to 1-year
corporates.
A strategy that uses futures that are in fact traded would require selling futures
today on 20-year government bonds. In one year, sell your corporate bonds and
use the proceeds to purchase an offsetting futures contract on 19-year
government bonds to close out your futures position. The additional risk with this
strategy is basis risk, which is the risk that the prices of government bonds and
government bond futures will not move in exactly the same way as corporate
bonds of the same maturity.
To lock in today's rates, sell $40 million of 10-year government bond futures. If
interest rates rise, the value of the futures will fall, which means a profit for you
since you are short the futures. At the end of three months, when your own bond
issue is floated, close out your futures position by buying an offsetting futures
contract. If interest rates have in fact risen, use the profits from your futures position
to finance the increased interest payments on your bond issue. If interest rates
have fallen, use some of the proceeds from your bond issue to cover your loss from
your futures position. Either way, ignoring basis risk, the effective interest rate on
your bond issue is locked in at today's rates by selling futures.
R A RF 14 3
1.833
VA 6
See the table in the answers to Problem 5 for the remaining funds Sharpe ratios.
The Treynor ratio is similar to the Sharpe ratio, except the funds beta is used in
the denominator instead of the standard deviation. E.g., for fund A we have:
R A RF 14 3
7.833
EA 1.5
See the table in the answers to Problem 5 for the remaining funds Treynor ratios.
A funds differential return, using standard deviation as the measure of risk, is the
funds average return minus the return on a nave portfolio, consisting of the
market portfolio and the riskless asset, with the same standard deviation of return
as the funds. E.g., for fund A we have:
R m RF 13 3
R A RF uV A 14 3 u 6 1%
Vm 5
See the table in the answers to Problem 5 for the remaining funds differential
returns based on standard deviation.
A funds differential return, using beta as the measure of risk, is the funds
average return minus the return on a nave portfolio, consisting of the market
portfolio and the riskless asset, with the same beta as the funds. This measure is
often called Jensens alpha. E.g., for fund A we have:
R A RF R m RF u E A 14 3 13 3 u 1.5 4%
See the table in the answers to Problem 5 for the remaining funds Jensen alphas.
This differential return measure is the same as the one used in Problem 4, except
that the riskless rate is replaced with the average return on a zero-beta asset. E.g.,
for fund A we have:
RA RZ Rm RZ u E A 14 4 13 4 u 1.5 3.5%
The answers to Problems 1 through 5 for all five funds are as follows:
Differential
Return Differential Differential
Based On Return Return
Sharpe Treynor Standard Based On Based On
Fund Ratio Ratio Deviation Beta and RF Beta and R Z
Looking at the table in the answers to Problem 5, we see that Fund B is ranked
higher than Fund A by their Sharpe ratios. Solving for the average return that
would make Fund Bs Sharpe ratio equal to Fund As we have:
R B Rf RB 3
1.833
VB 4
or
RB 10.33%
So, for the ranking to be reversed, Fund Bs average return would have to be
lower than 10.33%.
A.
The points on a Predictive Realization Diagram would have the following
coordinates (where Pi = predicted change in earnings and Ri = realized change in
earnings):
Industry/Firm Pi Ri
A1 0.05 0.00
A2 0.05 0.03
A3 0.75 0.25
B4 0.04 0.06
B5 0.05 0.04
B6 0.65 0.20
B7 0.01 0.01
C8 .070 0.40
C9 0.03 0.01
C10 0.02 0.02
While there are only ten points on the Prediction Realization Diagram, certain
tendencies can be detected. It is very clear from the diagram that analysts in this
brokerage firm systematically overestimate earnings. Their forecasts have a strong
upward bias. The second marked tendency is fro the degree of overestimation to
grow as positive increases in earnings become larger. Similarly, there is a slight
(based on one observation) tendency for analysts to overestimate the size of a
decrease in earnings when a decrease takes place. The analysts misestimated the
direction of a change in earnings in only two out of the ten cases.
B.
Recall from the text that, for the computation of mean square forecast error
(MSFE), the results are the same whether we use predicted levels or predicted
changes in earnings. We will do the MSFE analysis using levels and the following
formula:
N
1
MSFE
N
F A
i 1
i i
2
where Fi is the forecasted level of earnings for firm i per share Ai is the actual
earnings per share for firm i.
Sum 1.2979
Therefore:
1.2979
MSFE 0.1298
10
C.
From the text, we know that the MSFE can be decomposed by level of
aggregation as follows:
P R >P @ >P P R @
N N
2 1 2 1 2
MSFE a P Ra R i a i Ra
N i 1 N i 1
where the first term measures the forecast error due to all analysts misestimating
the average earnings in the economy, the second term measures the error due to
individual analysts misestimating the differential earnings for particular industries
from the average for the economy, and the third term measures the error due to
individual analysts misestimating the differential earnings for particular companies
within an industry from the average for that industry. So we have:
P R 2
0.223 0.0482 0.0306
>P P R @
N
1 2
i a i Ra 0.0849
N i 1
Notice that the sum of the three components equals 0.1298, which is the total
MSFE we calculated earlier.
D.
1. MSFE for each analyst:
3
1 1
P R
2
MSFE(A) u i i u 1.0029 0.3343
3 i 1 3
MSFE(B) = 0.0508
MSFE(C) = 0.0307
For analyst A,
Industry Error
= PA RA
2
= 0.1272
>P P R @
3
1 2
Company Error = i A i RA
3 i 1
3
=
1
>Pi 0.2833 Ri 0.0733@2
3 i1
= 0.2071
0.1272
% Industry Error = u 100 = 38.05%
0.3343
0.2071
% Company Error = u 100 = 61.95%
0.3343
For analyst B,
1 4
Company Error = >Pi 0.1825 Ri 0.0725@2
4 i1
= 0.0387
For analyst C,
1 3
Company Error = >Pi 0.2167 Ri 0.1367@2
3 i1
= 0.0243
0.0306
% Error Due To Bias = u 100 = 23.57%
0.1298
0.0248
% Error Due To Variance = u 100 = 19.11%
0.1298
0.0744
% Error Due To Covariance = u 100 = 57.32%
0.1298