Estimators of the disturbance variance in econometric models. Small- sample bias and the existence of moments (Q1117662): Difference between revisions

For general linear and non-linear models with stochastic regressors, we give simple exact bounds on the expected value of standard least-squares estimators of the disturbance variance. The bounds are valid for any correlation structure between the disturbances. We give simple conditions for residuals and variance estimators to have finite moments. In particular, for normal disturbances, all the moments exist. We also present analogous results for generalized least squares, simple and weighted \(L_ p\) estimation, and maximum likelihood. In the latter case, we give an information inequality related to the estimation of the entropy of a distribution.