Bias of LS estimators in nonlinear regression models with constraints. I: General case. (Q1775170): Difference between revisions

An asymptotic approximation of the bias of least squares estimators in nonlinear regression models with parameters which are subject to nonlinear equality constraints is derived. The normality of the observation vector is assumed, however under some conditions the results are valid for non-normal distribution as well. Different assumptions on constraints and on the model are taken into account. For functions of the parameters the invariance of the approximate bias with respect to reparametrization is demonstrated. Singular models are considered as well.