Bias Correction for Estimators of the Residual Variancein the ARMA ( 1 , 1 ) Model 1

We consider the ARMA(1,1) model and deal with the estimation of the residual variance. Results are known for the maximum likelihood(ML) es-timators under normality, both for known and unknowm mean, in which case the asymptotic biases depend on the number of parameters(including the mean) and on the true residual variance, but not on the values of the remaining parameters. For moment and least squares estimators the situation is diierent: the asymptotic biases depend on the values of the parameters, besides the true variance. Some simulation results are also presented.