SHRINKAGE TESTIMATORS OF SCALE PARAMETER FOR EXPONENTIAL MODEL UNDER ASYMMETRIC LOSS FUNCTION
The present paper proposes shrinkage testimator(s) for the scale parameter for an exponential distribution. An important feature of the proposed testimator is that, it removes the arbitrariness in the choice of shrinkage factor (weights) by making it dependent on the test statistic. The risk properties of the proposed testimator(s) have been studied under asymmetric loss function. It has been observed that the proposed testimator performs better than the classical Uniformly Minimum Variance Unbiased Estimator (UMVUE). Recommendations regarding its applications for various degrees of asymmetry (over/under estimator), level(s) of significance have been made.