|
Best Linear Minimum Bias Estimation in Linear Regression
Peter Schonfeld
Abstract
In this paper the problem of best (minimum variance) linear estimation of the regression coefficients in the univariate linear regression model is considered when no assumption about the rank of either the moment matrix of the regressors or the variance-covariance matrix of the error term is made. This problem has sometimes been called the "generalization of generalized least squares." Some brief examples where this problem is relevant in econometrics are given in the introductory part.
|
