M-Estimation of Scale Parameters in a Structural Time Series Model
We develop scale estimators of a structural time series model which are robust towards additive outliers. This is done by extending the application of the $M$-estimation technique to the scale estimation problem in time series data. A Monte Carlo experiment is carried out to study the robust properties of the proposed estimators. The simulation results indicate that the proposed $M$-estimators clearly outperform the maximum likelihood estimators produced by the Kalman filter when the observations are contaminated by outliers.
Journal of Applied Statistical Science
Chow, Hwee Kwan.
M-Estimation of Scale Parameters in a Structural Time Series Model. (1996). Journal of Applied Statistical Science. 3, (1), 93-105. Research Collection School Of Economics.
Available at: http://ink.library.smu.edu.sg/soe_research/462