(1) Institute of Statistical Science, Academia Sinica, 115 Taipei, Republic of China;(2) Department of Mechanics and Mathematics, National Taras Shevchenko University, Vladimirskaya st. 64, 01033 Kiev, Ukraine
Abstract:
The linear normal multivariate errors-in-variables model is considered. The model involves an equation error. It is proved in both structural and functional cases that the first moment of the adjusted least squares estimator does not exist.