EViews.com

hi

i have used Newey and West standard errors to correct serial correlation and heteroskadasticity for my regression model. Still when i perform LM test it reports the existence of second order autocorrelation. Why is this happening? Are NW std errors not sufficient to eliminate the serial correlation? Should i go for GLS method for my model. I am performing time series regression.

Plz suggest. Thanks

Newey-West standard errors do not have any effect on the existence of serial correlation. They just give you the correct standard errors when serial correlation is present.