Detection of additive outliers in bilinear time series
Computational Statistics & Data Analysis
Time Series Analysis, Forecasting and Control
Time Series Analysis, Forecasting and Control
Hi-index | 0.00 |
An outlier detection procedure for BL (1,1,1,1) model is developed based on the maxima of the test statistics measuring the effects of IO, AO, TC and LC. A simulation study is carried out in order to investigate the sampling properties of the maxima of the outlier test statistics. It is associated with the sample size, the type of outlier and the coefficients chosen for BL (1,1,1,1). The results show that, in general, the performance of the detection procedure is good. The outlier detection procedure performs well in detecting AO for large value of ωAO. As for IO, the performance of outlier detection procedure is better for model with larger coefficient values. The outlier detection procedure is capable of detecting TC and LC, though the performance is affected if ω is large.