Professional Documents
Culture Documents
J-lm, what is
2. Given the implied risk free return rm, what is the implied market
portfolio Xm or equivalently, what is J-lm or tm from which all other
relevant quantities can be deduced?
3. Given the slope R of the CML, what is the implied market port
folio Xm or equivalently, what is J-lm or tm from which all other
relevant quantities can be deduced?
59
60
Jlp
I m plied CML
I m plied risk
free return
a,
r
Chap. 4
Figure
4.1
112
0
J.lm , rm
4. 1
Sec. 4. 1
Direct Derivation
61
l'x
subject to :
>
1,
(4. 1 )
{32
f..Lp
a0 + (a1 ( aP - (30)) 2 ,
(4.2)
(4.3)
from which
df..Lp
dap
a1am
(a 1 ( a - (30) ) !
The tangent line to the efficient frontier at (am, J.Lm) is thus
a1am
f..Lp = f..Lm +
r-1 ) ) ! ( ap - am ) .
( al ( am - fJO 2
Lemma 4.1
62
Chap. 4
>
rm = Go
G t/3o
m
1-l - Go
Wl
'th t m _
1-lm - Go .
G1
(b) Let the risk free return be rm with rm < Go . Then the expected
return of the implied market portfolio is given by
1-lm = Go +
f3o
G d3o
with tm =
Go - rm
Go - rm
>
Proof:
( a ) rm is the intercept of the line specified in Lemma 4. 1 , with ap = 0.
Thus
( 4.4 )
G 1 a!
2
!
R )) 2
( G } ( am - f-'0
G 1 (a! - f3o ) - Gt a!
Go +
( Gt (a - f3o ) ) !
Gt f3o
Go m
1-l - Go
( GI ( am - f3o ) ) !2
Direct Derivation
Sec. 4 . 1
Noting that /-lm
63
/-lm
ao - rm '
a1
f3o
ao - rm
which verifies the desired result for 1-lm Recall that J.lm
Comparing this formula with ( 4.5) it follows that
(4.5)
=
a0 + a1tm .
f3o
ao - rm '
as required.
(c) From Lemma 4. 1 , the slope of the CML is
( a1 (a;, - f3o ))
But a;, = {30 + t;,/32 Equating the two expressions for a;, and using
the fact that /32 = a1 gives
/-lm
64
Chap. 4
We now use this expression for J.Lm in the expression for the implied
risk free return in part (a) . This gives
a1f3o
In Theorem 4. 1 (a) recall from (2. 12) , (2. 13) and (2. 17) that a1 > 0
and 0 > 0. It then follows that rm < a0 and that rm tends to ao as
J.Lm tends to infinity. See Figure 4 . 1 .
{3
J.L
0
0 0. 0006
0.0100 0.0007
0
0
0.0007 0.0500 0.0001
0
0
0 0.0001 0.0700
0
0
0 0.0800
0
0
0 0.0900
0.0006
0
The efficient set coefficients could be found by hand calculation. However, this would require inverting the ( 5, 5 ) covariance matrix which is
Sec. 4. 1
Direct Derivation
65
ho =
h1 =
-4.3919
0.2057
0.8181
1 .59 1 1
1 . 7770
and
1 . 1427
_ 0. 7180 X 0.0065
1 .4 - 1 . 1427
1 . 1246.
We also calculate
culate
f3o o: 1 = 1 .4,
O:o - Tm
as expected. Finally, we calculate O"; according to
f.lm = O:o +
O"
o: 1 /3o = 1 . 1930.
O:o - Tm
66
Chap. 4
Xm
0.3294
0. 1353
0 . 1 500
0. 1928
0 . 1 925
0
4.2
0
Figure
4.2
Sec. 4.2
Optimization Derivation
67
(f.-lp - rm)/ap
(am, f.-lm )
{ f.-l(x1X1x)2- rm
1
llx
(4.6)
f.-l1X rm
(4.6) is
(4.7)
1,
(4.8)
where u is the multiplier for the budget constraint. Note that the quan
tities within square brackets are scalars, whereas
and are vec
tors.
f.-l, x l
Now, recall the optimization problem (2.3) and its optimality con
ditions (2.4) . We repeat these for convenience:
m1"n{
1}
(4.9)
68
Chap. 4
(4. 10)
Comparing (4.8) with (4. 10) shows that the optimality conditions,
and thus the optimal solutions for ( 4.9) and ( 4.6) will be identical
provided
x'Ex
(4. 1 1 )
1-1X - rm '
or equivalently
-t' x -
x'Ex
(4. 12)
(a) For fixed rm, let x0 be optimal for (4. 6}. Then xo is
also optimal for (4. 9} with t = xExo /(-t'x o - rm)
Theorem 4.2
1-1Xl -
Theorems 4. 1 (b) and 4.2 (a) are closely related. Indeed, Theorem
4.2 (a) implies the results of Theorem 4 . 1 (b) as we now show.
Let rm be given and let x0 be optimal for (4.6) . Then Theorem 4.2
(a) asserts that x0 is optimal for (2.3) with
(4. 13)
Because x0 is on the efficient frontier for (2.3) we can write xEx0 =
/30 + f32 t 2 and -t'x0 = a0 + a 1 t . Substitution of these two expressions in
Optimization Derivation
Sec. 4.2
69
f3o
ao - rm '
which is the identical result obtained in Theorem 4. 1 ( b ) .
We can perform a similar analysis beginning with Theorem 4.2 ( b ) .
Set tm = t 1 and Xm = X 1 in Theorem 4.2 ( b ) . Because Xm is optimal
for ( 2.3 ) , J-lm = ao + tmal and 0";, = f3o + t;. f32 Then Theorem 4.2 ( b )
implies
(4 . 1 4)
tm =
f3o
ao - rm
and J-lm
ao +
f3o
a1 ,
ao - rm
(4 . 1 5 )
70
f- Jl
p
Rcr
p
Chap. 4
Figure
4.3
The line /-lp - Rap = has slope R and is parallel to the specified
CML. As is increased (or equivalently /-lp - Rap is increased) , the line
will eventually be identical with the CML. The optimization problem
to be solved is thus
(4. 16)
(4. 17)
Optimization Derivation
Sec. 4.2
71
J.l
x
1 = ul,
(x ' x) 2
(4. 18)
where u is the multiplier for the budget constraint. This can be rewritten as
[ l J.l
(x'x)
R
X _
(x'x)
u l.
R
(4. 19)
Comparing (4. 19) with (4. 10) shows that the optimality conditions,
and thus the optimal solutions for (4.9) and (4. 19) will be identical
provided
t
or equivalently
(x'x)
t
(4.20)
Theorems 4.1 (c) and 4.3(b) are closely related. Indeed, Theorem
4.3(b) implies the results of Theorem 4.1 (c) as we now show. Let R be
the given slope of the CML, assume
R > J1j;
( 4.21)
and let tm = t 1 in the statement of Theorem 4.3(b) . We can determine
tm and rm as follows. Theorem 4.3(b) asserts that
R =
(xxm ) l / 2
72
Chap. 4
{30 + t'!.f32 .
( 4.22 )
( 4.23 )
Sec. 4.3
73
R2
dMP
dP
a 1 p
(al( - f3o ) )
I
f3i
(4.24)
and
x(t1) x1
t1. (M'Xl - rm)/(xxl)
1
M1Xl - xxdt1,
M1X1 - R(xx
1)2
R (x x1) /t1.
4.3
>
74
Chap. 4
a {30 t{32 .
()
a;a; ( )
( )
a; a;0
()
Table 4.1 shows these pairs for a variety of values of t for the problem
of Example 2 . 1 . For example, for t = 0.10, the minimum portfolio vari
ance is 0.0134043 when the expected return is required to be 1 . 19574.
Conversely, the maximum expected return is 1 . 19574 when the variance
is held at 0.0134043.
TABLE
4.1
t
0. 1000
0. 1 1 1 1
0.1250
0.1429
0.1667
0.2000
0.2500
0.3333
0.5000
1.0000
2.1.
a;
t0
R(a1 2 ) 1 12
R (a) / /
rm
ajt0,
( ) rm
Computer Programs
Sec. 4.3
75
Table 4.2 shows these pairs for a variety of values of t for the problem
of Example 2. 1. For example, for t = 0. 10, the maximum Sharpe ratio is
1.15777 for a fixed risk free return of 1 .0617. Conversely, the maximum
risk free return is 1 .0617 for a specified Sharpe ratio of 1 . 15777.
TABLE
4.2
t
0.1000
0. 1 1 1 1
0.1250
0. 1429
0. 1667
0.2000
0.2500
0.3333
0.5000
1.0000
2. 1 .
Tm
Tm
Note that in Table 4.2 the two expressions for Sharpe ratios, namely
(/-lFurthermore,
R = (a-;0for) 112 /tothe arerisknumerically
identical.
pO - rm)/(a-)the112twoandexpressions
free return, namely
rm = /-lpo - a-;0 /to and rm = /-lpo - R(a-;0) 112 produce identical nu
merical results. The equality of these two pairs of expressions is true
in general. See Exercise 4.8.
4.4
Computer Programs
Figure 4.4 displays the routine Example4pl.m which performs the cal
culations for Example 4. 1 . The data is defined in lines 1-6 and line 8
invokes the routine EFMVcoeff ( see Figure 2.6 ) to compute the coeffi-
76
Chap. 4
cients for the efficient frontier. The computations for parts ( a) , (b) and
( c ) of Example 4 . 1 are then done in a straightforward manner.
1
2
3
4
6
8
9
% E x amp l e 4 p 1 . m
mu = [ 1 . 1 1 . 1 5 1 . 2 1 . 2 7 1 . 3 ] '
S i gma = [ 1 . e-2 0 . 7 e- 3 0 0 0 . 6 e - 3
0 . 7 e- 3 5 . O e-2 1 . e - 4 0 0 ;
0 1 . e- 4
7 . 0 e-2 0 0
0 0 0 8e-2 0 ;
0 . 6 e - 3 0 0 0 9 . e- 2 ]
c h e c k d at a ( S i gma 1 1 . e - 6 ) ;
[ a l p h a O 1 a lpha 1 1 b e t a O 1 b e t a 2 1 h O 1 h 1 ]
EFMVco e f f ( mu 1 S i gma ) ;
10
11
12
13
14
15
16
11
% E x amp l e 4 . 1 ( a )
s t r i n g = ' E xamp l e 4 . 1 ( a ) '
mum = 1 . 4
rm = a l ph a O - ( a lpha 1 * b e t a 0 ) / ( mum - a lpha O )
tm = b e t a O / ( a l pha O-rm )
mum = a lpha O + tm * a l ph a 1
s i gp 2 = b e t a O + t m * t m * b e t a 2
18
% E x amp l e 4 . 1 ( b )
s t r i n g = ' E xamp l e 4 . 1 ( b ) '
21 rm = 1 . 0 5 0 0 0
22 t m = b e t a O / ( a l p h a O - rm)
23 mum = a lp h a O + a lpha 1 * tm
24 s i gp 2 = b e t a O + t m * t m * be t a 2
25
26 rm = 1 . 0 5 5
2 7 t m = b e t a O / ( a lp h a O - rm )
28 mum = a lpha O + a l pha 1 * tm
29 s i gp 2 = bet a O + t m * t m * be t a 2
30
3 1 % E xamp l e 4 . 1 { c )
32 s t r i ng = ' E xamp l e 4 . 1 ( c ) '
33 rm = 1 . 0 5
34 t m = b e t a O / ( a lp h a O - rm)
35 port = h O + tm * h 1
19
20
Figure
4.4
Example4pl .m.
Sec. 4.3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
1s
16
11
18
19
20
21
22
23
24
25
26
Computer Programs
77
% S h a rpeRat i o s . m
mu = [ 1 . 1 1 . 2 1 . 3 ] '
S i gma = [ 1 . e-2 0 0 ; 0 S . O e- 2 0 ; 0 0 7 . 0 e- 2 ]
c h e c k dat a ( S i gma , 1 . e - 6 ) ;
EFMVco e f f ( mu , S i gma ) ;
[ a lph a O , a lpha 1 , bet a O , b e t a 2 , h O , h 1 ]
for i = 1 : 1 0
t = 1 / ( 1 1-i )
x1 = hO + t . *h1 ;
mup = mu ' * X 1 ;
s i gp 2 = x 1 ' * S i gma * x 1 ;
s t r = ' Mi n imum v a r i a n c e % g f o r f i x e d e xpe c t e d r e t u r n o f
%g\n ' ;
fpr i n t f ( s t r , s i gp 2 , mup )
s t r = ' Max imum expe c t e d r e t u r n % g f o r f i x e d v a r i a n c e o f
%g\n ' ;
fp r i n t f ( s t r , mup , s i gp 2 )
rm = mup - s i gp 2 / t ;
S r at i o = ( mup - rm ) / s i gp 2 ' ( 1 / 2 ) ;
fp r i nt f ( ' Maximum S h a rpe r a t i o i s % g f o r rm % g \ n ' ,
S r at i o , rm )
S r at i o = s i gp 2 ' ( 1 . / 2 . ) I t ;
rm = mup - S r at i o * s i gp 2 ' ( 1 / 2 ) ;
fp r i nt f ( ' Max r i s k f r e e r e t u r n i s % g f o r S h a rpe r a t i o
%g\n ' . . .
, rm, S r at i o )
end
Figure
4.5
SharpeRatios . m .
Chap. 4
78
4.5
Exercises
4.1 For the data of Example 2 . 1 , determine the implied risk free re
turn corresponding to (a) /-lp = 1 .2, (b) /-lp = 1 .25.
4.2 What market portfolio would give an implied risk free return of
rm = 0?
4.3 Suppose rm = 1 . 1 is the risk free return for the data of Example
2 . 1 . Find the implied market portfolio.
4.4 If we use the fact that the optimal solution must lie on the efficient
frontier, then (4.6) can be formulated as
1 . 13, o:1
0. 72,
a!.
Sec. 4.4
Exercises
79