A Cram\'er-Rao inequality for non differentiable models
Statistics Theory
2012-04-13 v1 Statistics Theory
Abstract
We compute a variance lower bound for unbiased estimators in specified statistical models. The construction of the bound is related to the original Cram\'er-Rao bound, although it does not require the differentiability of the model. Moreover, we show our efficiency bound to be always greater than the Cram\'er-Rao bound in smooth models, thus providing a sharper result.
Cite
@article{arxiv.1204.2763,
title = {A Cram\'er-Rao inequality for non differentiable models},
author = {Thibault Espinasse and Paul Rochet},
journal= {arXiv preprint arXiv:1204.2763},
year = {2012}
}