25.06.2013 Views

statistique, théorie et gestion de portefeuille - Docs at ISFA

statistique, théorie et gestion de portefeuille - Docs at ISFA

statistique, théorie et gestion de portefeuille - Docs at ISFA

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

460 14. Gestion <strong>de</strong> Portefeuilles multimoments <strong>et</strong> équilibre <strong>de</strong> marché<br />

E Calcul<strong>at</strong>ion of the moments of the distribution of portfolio r<strong>et</strong>urns<br />

L<strong>et</strong> us start with equ<strong>at</strong>ion (72) in the 2-ass<strong>et</strong> case :<br />

ˆPS(k)<br />

1<br />

=<br />

2π 1 − ρ2 <br />

<br />

dy1dy2 exp − 1<br />

2 ytV −1 <br />

<br />

y1 <br />

y + ik χ1w1sgn(y1) <br />

√<br />

<br />

2<br />

<br />

<br />

y2 <br />

+χ2w2sgn(y2) <br />

√<br />

<br />

2<br />

<br />

Expanding the exponential and using the <strong>de</strong>finition (67) of moments, we g<strong>et</strong><br />

Posing<br />

1<br />

Mn =<br />

2π 1 − ρ2 <br />

dy1 dy2<br />

γq1q2 (n, p) = χ1 p χ2 n−p<br />

this leads to<br />

2π 1 − ρ 2<br />

<br />

n<br />

p=0<br />

<br />

n<br />

χ<br />

p<br />

p<br />

dy1dy2 sgn(y1) p<br />

<br />

y1 <br />

<br />

√<br />

<br />

2<br />

<br />

Mn =<br />

1χn−p 2 wp 1wn−p 2<br />

×sgn(y2) n−p<br />

q1p<br />

<br />

y2 <br />

<br />

√<br />

<br />

2<br />

<br />

sgn(y1) p<br />

<br />

y1 <br />

<br />

√<br />

<br />

2<br />

<br />

q2(n−p)<br />

sgn(y2) n−p<br />

<br />

y2 <br />

<br />

√<br />

<br />

2<br />

<br />

L<strong>et</strong> us <strong>de</strong>fined the auxiliary variables α and β such th<strong>at</strong><br />

α = (V −1 )11 = (V −1 )22 = 1<br />

1−ρ 2 ,<br />

β = −(V −1 )12 = −(V −1 )21 = ρ<br />

1−ρ 2 .<br />

n<br />

p=0<br />

q1<br />

+<br />

q2 <br />

q1p<br />

×<br />

. (143)<br />

1<br />

−<br />

e 2 ytV −1y . (144)<br />

q2(n−p)<br />

1<br />

−<br />

e 2 ytV −1y , (145)<br />

<br />

n<br />

w<br />

p<br />

p<br />

1wn−pγq1q2 2 (n, p) . (146)<br />

(147)<br />

Performing a simple change of variable in (145), we can transform the integr<strong>at</strong>ion such th<strong>at</strong> it is <strong>de</strong>fined<br />

solely within the first quadrant (y1 ≥ 0, y2 ≥ 0), namely<br />

γq1q2 (n, p) = χ1 p n−p 1 + (−1)n<br />

χ2<br />

2π 1 − ρ2 +∞ +∞<br />

dy1 dy2<br />

0<br />

0<br />

q1p q2(n−p) y1 y2<br />

α<br />

− √2 √2 e 2 (y2 1 +y2 2 ) ×<br />

<br />

× e βy1y2<br />

<br />

p −βy1y2<br />

+ (−1) e<br />

. (148)<br />

This equ<strong>at</strong>ion imposes th<strong>at</strong> the coefficients γ vanish if n is odd. This leads to the vanishing of the moments<br />

of odd or<strong>de</strong>rs, as expected for a symm<strong>et</strong>ric distribution. Then, we expand e βy1y2 + (−1) p e −βy1y2 in series.<br />

Permuting the sum sign and the integral allows us to <strong>de</strong>couple the integr<strong>at</strong>ions over the two variables y1 and<br />

y2:<br />

γq1q2 (n, p) = χ1 p n−p 1 + (−1)n<br />

χ2<br />

2π 1 − ρ2 +∞<br />

[1 + (−1)<br />

s=0<br />

p+s ] βs<br />

s!<br />

36<br />

+∞<br />

dy1<br />

0<br />

+∞<br />

× dy2<br />

0<br />

y q1p+s<br />

1<br />

2 q1p 2<br />

y q2(n−p)+s<br />

2<br />

2 q2 (n−p)<br />

2<br />

α<br />

−<br />

e 2 y2 1<br />

<br />

α<br />

−<br />

e 2 y2 1<br />

<br />

×<br />

. (149)

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!