Relationship Between an Innovative Outlier Model and the Intervention Effects Model

We present a relationship between an alternative estimation procedure of ARMA models based on an innovative outlier framework and the Kalman Filter Smoothers (KFS) estimation of intervention effects for outliers generated by special events. The intervention model describes how the events manifest themselves in the observations. The results are asymptotically equivalent to the estimation of ARMA parameters based on a generalized t-distribution for the residuals with a mixture interpretation relying on the innovative outlier framework. The joint estimation of the estimators, if considered, supports the normality test of the residuals.