Abstract:
The risk performances, under the symmetric squared error loss function, of the estimators
of the regression coefficients after a preliminary test for serial correlation have been widely
investigated in the literature. However, it is well known that the use of the symmetric loss functions
is inappropriate in estimation problems where underestimation and overestimation have different
consequences. We consider the Linear Exponential and Bounded Linear Exponential loss functions
which allows for asymmetry. The risks of the estimators are derived and numerically evaluated by
using simulations.
