Outliers in multivariate time series

This paper generalises four types of disturbance commonly used in univariat e time series analysis to the multivariate case, highlights the differences between univariate and multivariate outliers, and investigates dynamic eff ects of a multivariate outlier on individual components. The effect of a mu ltivariate outlier depends not only on its size and the underlying model, b ut also on the interaction between the size and the dynamic structure of th e model. The latter factor does not appear in the univariate case. A multiv ariate outlier can introduce various types of outlier for the marginal comp onent models. By comparing and contrasting results of univariate and multiv ariate outlier detections, one can gain insights into the characteristics o f an outlier. We use real examples to demonstrate the proposed analysis.