COMPLETE CONVERGENCE OF ERROR VARIANCE ESTIMATES UNDER PHI-MIXING ERROR IN LINEAR-MODELS

In this paper, we consider the estimates sigma(n)2 of error variance sigma2 = V ar(e(i)) in the linear models Y(i) = x' (i)beta+e(i)(i = 1,2,...). We study the complete convergence of sigma(n)2 - sigma2 when the error {e(i)} is a sequence of identically distributed psi- mixing variables. And we also obtain the better convergence rates when {e(i)} is not identically distribution.