The standard errors of the parameter estimates of a fitted regression modelare likely to be underestimated if there is positive serial correlation in thedata. This implies that explanatory variables...

The standard errors of the parameter estimates of a fitted regression modelare likely to be underestimated if there is positive serial correlation in thedata. This implies that explanatory variables may appear as ‘significant’when they should not. Use GLS to check the significance of the variablesof the fitted model from§5.6.3. Use an appropriate estimate of the lag 1autocorrelation withingls. Once you use a GLS model then use an ARIMA model.