THÈSE Estimation, validation et identification des modèles ARMA ...
THÈSE Estimation, validation et identification des modèles ARMA ...
THÈSE Estimation, validation et identification des modèles ARMA ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Chapitre 4. Multivariate portmanteau test for weak structural V<strong>ARMA</strong> models 128<br />
<br />
|i−j| 1 ρ<br />
L<strong>et</strong>, Σij =<br />
ρ |i−j| <br />
the covariance matrix of the vector (ηit,ηj<br />
1<br />
t) ′ . A joint<br />
probability density function of the bivariate random vector (ηit,ηj t) ′ with mean vector<br />
zero is of the form<br />
f(x) = (2π) −1 |Σij| −1/2 exp[(−1/2)x ′ Σ −1<br />
ij x],<br />
with |Σij| = 1−ρ 2|i−j| . If |ρ| < 1 the matrix Σij is invertible and<br />
Σ −1<br />
ij =<br />
<br />
1 1 −ρ<br />
×<br />
1−ρ 2|i−j| |i−j|<br />
−ρ |i−j| <br />
.<br />
1<br />
Then we have<br />
f(x1,x2) =<br />
1<br />
2π <br />
−1<br />
exp<br />
1−ρ 2|i−j| 2(1−ρ 2|i−j| <br />
2<br />
x1 +x<br />
)<br />
2 2 −2ρ|i−j| <br />
x1x2<br />
<br />
.<br />
L<strong>et</strong>ting β = 1/2π 1−ρ 2|i−j| and α = 1−ρ 2|i−j| , we calculate the integral<br />
Ex 2 1 x2 2<br />
<br />
= β<br />
R 2<br />
x 2 1x22 e<br />
x1 −1 −ρ<br />
2<br />
|i−j| 2 x2 +x α<br />
2 <br />
2<br />
dx1dx2.<br />
S<strong>et</strong>ting u = x1 −ρ |i−j| <br />
x2 /α and s = x2, the Jacobian of this transformation is |J| =<br />
α. Then we have<br />
Ex 2 1x 2 <br />
2 = βα<br />
R2 2 −1<br />
|i−j| 2<br />
αu+ρ s s e 2 (u2 +s2 ) duds,<br />
<br />
= βα s 2 e −1<br />
2 s2<br />
<br />
<br />
2 2 |i−j| 2|i−j| 2<br />
α u +2αρ us+ρ s e −1<br />
2 u2<br />
<br />
du ds.<br />
R<br />
Integrating by parts and passing to polar coordinates, we have<br />
Ex 2 1x 2 <br />
2 = βα α 2√ 2π +ρ 2|i−j|√ 2πs 2<br />
<br />
Thus, we have<br />
then, we have<br />
R<br />
R<br />
= βα 2πα 2 +6πρ 2|i−j|<br />
= 1+2ρ 2|i−j| .<br />
s 2 e −1<br />
2 s2<br />
ds,<br />
Cov ǫ 2 it,ǫ 2 2<br />
j t = Eǫitǫ 2 j t − Eǫ 2 2 2 2<br />
it Eǫj t<br />
= E<br />
2k<br />
η 2 it−mη2 j t−m −1<br />
m=0<br />
= 1+2ρ 2|i−j|2k+1 −1,<br />
Cov ǫ 2 it ,ǫ2 <br />
2|i−j|<br />
j t−h = 1+2ρ 2k−h+1<br />
−1.