New Functional Estimator in Quadratic Errors-In-Variables Model

Elena Usoltseva, Alexander Kukush
Theory of Stochastic Processes
Vol.16 (32), no.2, 2010, pp.126-131
A quadratic structural errors-in-variables model is considered. Functional estimators that are generated by estimating the functions conditionally unbiased given the latent variable are studied. Those estimators are constructed without the knowledge of the latent variable distribution. A problem is studied how to construct an estimator from the class which has the smallest, in certain sense, asymptotic covariance matrix.