A mixed model for regressions

This paper gives a selt-contained analysis of a mixed model for regressions. The model differs from the ordinary covariance model as well as from the error components regression model considered by Mundlak (1963), Wallace & Hussain (1969) and some other writers. A single-equation mixed regression model is specified and its connexion with Spjøtvoll's mixed model (1966) for analysis of variance is examined . An different two-stage method for estimating the parameters of the single-equation model is also given. The multi-regression mixed model is defined and system estimators for the parameters of the model are derived, which have some optimal asymptetic properties.