A review of Brown 1971 (in)admissibility results under scale mixtures of Gaussian priors

Brown's 1971 paper “Admissible estimators, recurrent diffusions and insoluble boundary value problems” is a landmark in the admissibility literature. It nearly completely settles the issue of admissibility/inadmissibility for estimating the mean of a multivariate normal distribution with identity covariance under sum of squared error loss. We revisit this wonderful tour de force on its 50th anniversary and present an alternative and more direct proof of the result for generalized Bayes estimators corresponding to priors which are a subclass of scale mixtures of spherical normals.