Maximum likelihood estimation of the linear model with equicorrelated errors

Abstract

We provide a simple proof that the maximum-likelihood estimator in the linear model with equicorrelated errors does not exist, and explain why not.