Asymptotic Efficient Estimation of the Change Point in Time Series Regression Models

This paper discusses the problem of estimating unknown change point in the trend function of a time series regression model. The error process considered here is a Gaussian stationary process with spectral density. The asymptotic properties of quasi maximum likelihood (QMLE) and quasi Bayes (QBE) estimators are studied. Consistency, limiting distributions and convergence of higher order moments of the estimators are obtained. It is also shown that the QBE is asymptotically efficient, and that the QMLE is not so general.