On the existence of consistent estimates and tests