On asymptotic ancillarity and inference for Yule and regular nonergodic processes

Abstract

By extending the notion of ancillarity to asymptotic ancillarity a conditional inference approach to statistical inference for a class of regular nonergodic processes is developed. This approach is closely related to that of Efron & Hinkley (1978) for the independent identically distributed case. The Yule process is used to illustrate the main ideas.