On one conjecture of R. S. Singh (Q792036)

In the problem to obtain asymptotically optimal empirical Bayes estimators, speeds and the best possible speed at which the estimators are asymptotically optimal are needed. \textit{R. S. Singh} [Ann. Stat. 7, 890-902 (1979; Zbl 0411.62019)] conjectured that a rate \(O(n^{-1})\) or better is not possible for any empirical Bayes estimator in any Lebesgue- exponential family even though \(\Theta(\subset R^ 1)\) is bounded, when we use the square error loss function. In this paper it is shown that the weaker part of Singh's conjecture, the \(o(n^{-1})\) part, is correct. It remains unknown whether the \(O(n^{-1})\) part of the conjecture is correct or not.