DIAGNOSING SHOCKS IN TIME-SERIES

Efficient means of modeling aberrant behavior in times series are deve loped. Our methods are based on state-space forms and allow test stati stics for various interventions to be computed from a single run of th e Kalman filter smoother. The approach encompasses existing detection methodologies. Departures commonly observed in practice, such as outly ing values, level shifts, and switches, are readily dealt with. New di agnostic statistics are proposed. Implications for structural models, autoregressive integrated moving average models, and models with expla natory variables are given.