17.08.2013 Views

THÈSE Estimation, validation et identification des modèles ARMA ...

THÈSE Estimation, validation et identification des modèles ARMA ...

THÈSE Estimation, validation et identification des modèles ARMA ...

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.

Chapitre 3. Estimating the asymptotic variance of LSE of weak V<strong>ARMA</strong> models 108<br />

Lemma 3.9. Under Assumptions A1–A8, we have<br />

<br />

<br />

sup<br />

i,j<br />

ˆ <br />

<br />

Γn(i,j)−Γ(i,j) → 0 in probability as n → ∞.<br />

Proof of Lemma 3.9. For any θ ∈ Θ, l<strong>et</strong><br />

Mn ij,h(θ) := 1<br />

n<br />

By the ergodic theorem, we have<br />

n−|h| <br />

t=1<br />

e ′ t−h(θ)⊗ Id 2 (p+q) ⊗e ′ t−j−h(θ) <br />

⊗ e ′ t (θ)⊗ I d 2 (p+q) ⊗e ′ t−i (θ) .<br />

Mn ij,h(θ) → Mij,h(θ) := E e ′ t−h (θ)⊗ I d 2 (p+q) ⊗e ′ t−j−h (θ)<br />

⊗ e ′ t(θ)⊗ Id 2 (p+q) ⊗e ′ t−i(θ) a.s.<br />

A Taylor expansion of vec ˆ Mn ij,h around θ0 and (3.2) give<br />

vec ˆ Mn ij,h = vecMn ij,h +<br />

In view of (3.19), we then deduce that<br />

lim<br />

n→∞ sup<br />

<br />

<br />

sup <br />

<br />

i,j≥1 |h|

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

Saved successfully!

Ooh no, something went wrong!