| Article ID: | iaor20081523 |
| Country: | United Kingdom |
| Volume: | 25 |
| Issue: | 1 |
| Start Page Number: | 37 |
| End Page Number: | 47 |
| Publication Date: | Jan 2006 |
| Journal: | International Journal of Forecasting |
| Authors: | Giordani Paolo |
Chan and Cheung propose a GM approach to outlier robust estimation of threshold models. We show that their estimator can be inconsistent and extremely inefficient even when the model is correctly specified and the disturbances are normally distributed, and outline situations in which the problem is likely to be more severe.