A two-step estimation of a linear regression model with first-order serial correlation in the errors. In the first step the firstorder autocorrelation coefficient is estimated using the ordinary least squares residuals from the main regression equation. In the second step this estimate is used to rescale the variables so that the regression in terms of rescaled variables has no serial correlation in the errors. This is an example of feasible generalized least squares estimation.
|Reference: Oxford Press Dictonary of Economics, 5th edt.|