Correcting the influence of autocorrelated errors in linear regression models

The paper presents the case often met in a regression model, the autocorrelated errors. In the first part of the paper are summarized some theoretical issues about the sources of appearance of autocorrelated errors, some statistic tests to identify the autocorrelation and there are presented in more detail three alternatives of the classical methods for estimating parameters, methods that are better suited to the given situation: the Cochrane-Orcutt method (with its variant Yule-Walker method), the Durbin method and the Hildreth-Lu method. The second part of the paper presents an example of a regression model with autocorrelated errors and uses a method for correcting the influence of the autocorrelation on the estimated parameters, using the statistical package SAS 9.1.