The stability of the problem of statistical estimation and a choice of the loss function

In the article there is given the definition of the stability of the problem of statistical estimation. We prove that if the set of all symmetrical statistics form a complete class tnen the problem of statistical estimation is non-stable when the loss function of the form $W(\gamma^*,\gamma)=V(\gamma^*-\gamma)$ is used. Also there are given some examples of loss functions for which the.problem of statistical estimation is stable.