Robust unit root tests for autoregressive models : Notas de Matemática, 58

In this paper a robust test is developed for detecting a unit root for autoregressive models. The basic idea consists of replacing the least squares estimators in the Dickey-Fuller statistics by robust estimators with a high breakdown point and high efficiency called τ-estimators. The limiting distr...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autor principal: Ferretti, Nélida Elena
Formato: Publicacion seriada
Lenguaje:Español
Publicado: 1996
Materias:
Acceso en línea:http://sedici.unlp.edu.ar/handle/10915/170676
Aporte de:
Descripción
Sumario:In this paper a robust test is developed for detecting a unit root for autoregressive models. The basic idea consists of replacing the least squares estimators in the Dickey-Fuller statistics by robust estimators with a high breakdown point and high efficiency called τ-estimators. The limiting distribution of the test statistics proposed are obtained under the unit root null hypothesis. A Monte Carlo study is described, illustring the asymptotic efficiency of the τ-estimators and empirical power comparisons using moderate and large size samples for first-order autoregressive processes. The new tests are shown to have the desirable robust properties.