Shrinkage estimators of the location parameter for certain spherically symmetric distributions (Q1335359)

The shrinkage estimators of location parameters for normal distributions have been considered extensively in the statistical literature. Under quadratic loss, some conditions under which \(X + cg (X)\) dominates \(X\) were discussed by \textit{A. C. Brandwein} and \textit{W. E. Strawderman} [Ann. Stat. 19, No. 3, 1639-1650 (1991; Zbl 0741.62058)] when \(X\) has a spherically symmetric distribution, where \(c\) is a constant and \(g(\cdot)\) is a real function. This paper is devoted to improving the bounds on the constant \(c\) given by Brandwein and Strawderman. The investigation shows that the upper bounds of the constant \(c\) can be enlarged significantly for a subclass of spherically symmetric distributions. Several classical shrinkage estimators, such as James-Stein estimator and Brandwein-Strawderman estimator, are included in the class of the estimators with form \(X+cg(X)\). In addition, analogous results for quadratic loss are obtained for concave functions of quadratic loss and for general quadratic loss.