Professional Documents
Culture Documents
Outline
Literature and Introduction
Inverse Problem of Mean-Variance Formulation
Pseudo Efficiency
Type 1 and Best Investment Performance
Type 2 and Optimal Management of Initial Endowment
Conclusions
1 / 34
Outline
Literature and Introduction
Inverse Problem of Mean-Variance Formulation
Pseudo Efficiency
Type 1 and Best Investment Performance
Type 2 and Optimal Management of Initial Endowment
Conclusions
1 / 34
Outline
Literature and Introduction
Inverse Problem of Mean-Variance Formulation
Pseudo Efficiency
Type 1 and Best Investment Performance
Type 2 and Optimal Management of Initial Endowment
Conclusions
1 / 34
Outline
Literature and Introduction
Inverse Problem of Mean-Variance Formulation
Pseudo Efficiency
Type 1 and Best Investment Performance
Type 2 and Optimal Management of Initial Endowment
Conclusions
1 / 34
min
x
s.t.
1 !
x Vx
2
x ! e = ,
x ! 1 = x0 ,
(1)
1 !
x Vx + 1 (x ! e ) + 2 (x ! 1 x0 )
2
Optimality conditions:
Vx + 1 e + 2 1 = 0, x = V 1 (1 e + 2 1)
B
A
x ! e = , e ! V 1 e1 1! V 1 e2 =
x ! 1 = x0 , e ! V 1 11 1! V 1 12 = x0 .
"!
" !
"
!
"
!
"!
"
1 C
A
1
1
A
=
C
2
x0
2
A B
x0
D
4 / 34
(2)
where
A = 1! V 1 e = e ! V 1 1,
B = e ! V 1 e > 0,
C = 1! V 1 1 > 0,
D = BC A2 > 0.
The positiveness of D can be seen from the positiveness of
(Ae B1)! V 1 (Ae B1) = BD (Merton(1972)).
A fact that has not been fully recognized in the literature is that
parameter A can be positive, negative or zero.
5 / 34
= x ! (x0 ; )Vx(x0 ; )
x0
6 / 34
A
x0
C
$2
x02
.
C
(3)
A
x0 },
C
A
= {x(x0 ; ) | x(x0 ; ) is given in (2), < x0 }.
C
= {x(x0 ; ) | x(x0 ; ) is given in (2),
8 / 34
Reachable Region
The reachable region in the mean-variance space is defined as
{(, ) | | | B}.
!
Reachable Region
The reachable region in the mean-variance space is defined as
{(, ) | | | B}.
!
Reachable Region
The reachable region in the mean-variance space is defined as
{(, ) | | | B}.
!
Reachable Region
The reachable region in the mean-variance space is defined as
{(, ) | | | B}.
!
Part A
Part B
Part C
|A|
}
C
} represents
with positive expected returns and {(, ) | = |A|
C
all minimum variance points with negative expected returns.
12 / 34
Part A:
Part B:
|A|
{(, ) | < B}.
C
Every point in Part A is achieved by two efficient boundary portfolio
policies corresponding to two different initial wealth levels.
|A|
|A|
{(, ) | < },
C
C
Any point in Part B is achieved by one efficient boundary portfolio
policy and one inefficient boundary portfolio policy corresponding to
two different initial wealth levels. When A = 0, part B vanishes.
Part C:
|A|
{(, ) | B < < }.
C
Every point in Part C is achieved by two inefficient boundary
portfolio policies.
13 / 34
Part A:
Part B:
|A|
{(, ) | < B}.
C
Every point in Part A is achieved by two efficient boundary portfolio
policies corresponding to two different initial wealth levels.
|A|
|A|
{(, ) | < },
C
C
Any point in Part B is achieved by one efficient boundary portfolio
policy and one inefficient boundary portfolio policy corresponding to
two different initial wealth levels. When A = 0, part B vanishes.
Part C:
|A|
{(, ) | B < < }.
C
Every point in Part C is achieved by two inefficient boundary
portfolio policies.
13 / 34
Part A:
Part B:
|A|
{(, ) | < B}.
C
Every point in Part A is achieved by two efficient boundary portfolio
policies corresponding to two different initial wealth levels.
|A|
|A|
{(, ) | < },
C
C
Any point in Part B is achieved by one efficient boundary portfolio
policy and one inefficient boundary portfolio policy corresponding to
two different initial wealth levels. When A = 0, part B vanishes.
Part C:
|A|
{(, ) | B < < }.
C
Every point in Part C is achieved by two inefficient boundary
portfolio policies.
13 / 34
Pseudo Efficiency
Type 1 and Best Investment Performance
Outline
Literature and Introduction
Inverse Problem of Mean-Variance Formulation
Pseudo Efficiency
Type 1 and Best Investment Performance
Type 2 and Optimal Management of Initial Endowment
Conclusions
14 / 34
Pseudo Efficiency
Type 1 and Best Investment Performance
Definition
If an efficient mean-variance pair of problem (MV ) associated with initial
endowment x0 can be also generated or even dominated by another
mean-variance pair generated by another boundary portfolio policy
associated with initial wealth x0 which is strictly less than x0 , i.e.,
C
(, [
D
x0 < x0 ,
A
x0
C
$2
x2
C
+ 0 ]) ' (
, [
C
D
#
$2
A
x2
x0 + 0 ]),
C
C
(6)
(7)
Pseudo Efficiency
Type 1 and Best Investment Performance
2A
B
(8)
Pseudo Efficiency
Type 1 and Best Investment Performance
Remark
Set { |
A
C x0
B
A x0 }
is non-empty as D = BC A2 > 0.
17 / 34
Pseudo Efficiency
Type 1 and Best Investment Performance
Markowitzs Example
For this market of three risky assets with expected return vector
e = (1.162, 1.246, 1.228)!
and covariance,
18 / 34
Pseudo Efficiency
Type 1 and Best Investment Performance
From (8) and Proposition 3.2, it can be verified that the following
boundary policy associated with a less initial wealth x0 = 0.9985,
x(
x0 = 0.9985, = 1.16) = (0.9914, 0.0404, 0.0475)!
yields the same mean-variance pair of (1.160, 0.0719).
Pseudo Efficiency
Type 1 and Best Investment Performance
min
x
s.t.
1 !
x Vx
2
x ! e = ,
x ! 1 x0 ,
(9)
where the initial wealth level, x0 , and the pre-set expected wealth level,
, are both assumed to be positive.
20 / 34
Pseudo Efficiency
Type 1 and Best Investment Performance
x0
(10)
Pseudo Efficiency
Type 1 and Best Investment Performance
Pseudo Efficiency
Type 1 and Best Investment Performance
Pseudo Efficiency
Type 1 and Best Investment Performance
Pseudo Efficiency
Type 1 and Best Investment Performance
MVP
Pseudo Efficiency
Type 2 and Optimal Management of Initial Endowment
Outline
Literature and Introduction
Inverse Problem of Mean-Variance Formulation
Pseudo Efficiency
Type 1 and Best Investment Performance
Type 2 and Optimal Management of Initial Endowment
Conclusions
24 / 34
Pseudo Efficiency
Type 2 and Optimal Management of Initial Endowment
max
x
s.t.
x ! e + (x0 x ! 1)
1 !
x Vx = 2 ,
2
x ! 1 x0 ,
(14)
25 / 34
Pseudo Efficiency
Type 2 and Optimal Management of Initial Endowment
2
1
1
x (
x0 ; ) = (BV 1AV e)+
2 +
x0 (CV 1 eAV 1 1),
D
CD
C D
CD
(15)
where
$
x0 ,
if A > C and 2 > BA x0 ,
%
x0 =
(16)
C
(A C ) D+(AC )2 , otherwise.
Furthermore, the mean-variance efficient frontier of (MV2 ) can be expressed as,
%
)2 2
D 2
26 / 34
Pseudo Efficiency
Type 2 and Optimal Management of Initial Endowment
MVP
Pseudo Efficiency
Type 2 and Optimal Management of Initial Endowment
MVP
MVP
(b) A < 0
Pseudo Efficiency
Type 2 and Optimal Management of Initial Endowment
29 / 34
Pseudo Efficiency
Type 2 and Optimal Management of Initial Endowment
Pseudo Efficiency
Type 2 and Optimal Management of Initial Endowment
Markowitzs Example
Applying the two revised mean variance formulations, (MV1 ) and (MV2 ),
to the Markowitzs example yields the efficient frontier of (MV1 ) as
.
0.0554 2 0.0079 + 1.1540, > 1.1609,
=
3.0589,
0 1.1609,
and the efficient frontier of (MV2 ) as
.
0.0554 2 0.0079 + 1.1540, > 1.2055,
=
1 + 0.4702,
0 1.2055.
It can be seen that the mean variance pairs in
{(, 18.0388( 1.1540)2) + 0.1432) | 1.1540 1.1609} are pseudo
efficient (type 1) and the mean variance pairs in
{(, 18.0388( 1.1540)2) + 0.1432) | 1.1609 1.2055} are pseudo
efficient (type 2).
31 / 34
Pseudo Efficiency
Type 2 and Optimal Management of Initial Endowment
1.22
1.21
1.2
1.19
1.18
1.17
1.16
MVP
1.15
1.14
1.13
0.26
0.28
0.3
0.32
0.34
0.36
0.38
0.4
0.42
0.44
0.46
Conclusions
Conclusion
!
33 / 34
Conclusions
34 / 34