Asymptotically efficient estimation in the presence of heteroskedasticity of unknown form. (English) Zbl 0651.62107

This paper presents a generalization on the (- asymptotic) properties of weighted least squares estimators, where the residual variance is specified to be an unknown function of the explanatory variables. The paper is well written, short and concise; yet it manages to combine existing literature with own results, where the own results are densely concentrated in an appendix.
Heteroskedasticity has been one of the original topics leading to econometrics, where the earlier work used rather heroic assumptions concerning the specification of the kind of heteroskedasticity. One can only hope that possible generalizations of this specification - as indicated in this paper - will find wider acceptance in empirical econometrics.
Reviewer: G.Hasenkamp


62P20 Applications of statistics to economics
62G20 Asymptotic properties of nonparametric inference
62E20 Asymptotic distribution theory in statistics
62F12 Asymptotic properties of parametric estimators
62J05 Linear regression; mixed models
Full Text: DOI