We show under very parsimonious assumptions that FGLS and GLS are asymptotically equivalent when errors follow an invertible MA(1) process. Although the linear regression model with MA errors has been studied for many ...