Partenaires

CNRS
Logo tutelle
Logo tutelle
Logo tutelle


Rechercher

Sur ce site

Sur le Web du CNRS


Accueil du site > Séminaires > Probabilités Statistiques et réseaux de neurones > Testing covariance stationarity for high frequencies panel data : a nonparametric approach based on the evolutionary spectral density

Vendredi 3 juin 2005, à 10h

Testing covariance stationarity for high frequencies panel data : a nonparametric approach based on the evolutionary spectral density

Ibrahim Ahamada (Eurequa, Université Paris 1)

Résumé : This paper proposes a nonparametric test of covariance stationarity for high frequencies panel data. The Alternative hypothesis is the existence of at least an individual series which present a time varying covariance structure. Our approach is based on the concept of time varyng spectral density. Simulaton experiments confirm the efficiency of our approach. We show how our test detects diverse types of instabilities that classical test can do. An application to a panel data of financial returns reveals some instabilities in the covariance structure. This result confirm some recent works and call to take some precautions before using traditional stationarity tools to describe the behavior of some long sizes financial data.

Dans la même rubrique :