Inferences for the Linear Errors-in-Variables with Changepoint Models

Abstract

A new linear structural errors-in-variables regression with changepoint model is considered. In this model we consider the likelihood ratio test based on the maximum Hotelling T ² for the test of no change against the alternative of exactly one change. If there is a change, either known a priori or by testing, then we estimate the unknown changepoint parameter and some other related parameters by maximum likelihood. The limiting distribution of the changepoint estimator is investigated and it is shown that the asymptotic efficiency increases as the absolute regression slope coefficient increases. A Monte Carlo study shows that the proposed estimator performs satisfactorily.

Key Words:

• Asymptotic distribution
• Asymptotic efficiency
• Consistency
• Likelihood ratio test