Robust estimation for vector autoregressive models

NORA MULER; VICTOR J. YOHAI

COMPUTATIONAL STATISTICS AND DATA ANALYSIS

ELSEVIER SCIENCE BV

Lugar: Amsterdam; Año: 2012

0167-9473

In this paper we extend the MM-estimators based on the boundedninnovation propagation AR model to VAR models .These estimators have a filtering mechanism that avoids the propagation of the effect of one outlier to the residuals of the subsequent periods . Besides, they are consistent and have the same asymptotic normal distribution than regular MM-estimators for VAR models. We perform a Monte Carlo study which shows that these estimators compare favorable with respect to other robust estimators.