Abstract: | This paper introduces a new approximate GLS estimator for the regression model with MA(1) disturbances, which outperforms both the Balestra (1980) and the Park and Heikes (1983) approximations and which is almost as efficient as the exact GLS estimator. |