Page 1 of 1

Autocorrelation

Posted: Wed May 23, 2012 2:02 pm
by yoyoo
Hi,
How can I eliminate the autocorrelation to reach Durbin Watson close to 2?
What are the steps to do that?

Thank you in advance.

Re: Autocorrelation

Posted: Wed May 23, 2012 3:13 pm
by startz
Look at the AR(1) command in the help system.

Re: Autocorrelation

Posted: Fri May 25, 2012 8:02 am
by yoyoo
Hi,

I followed the instructions but a message appeared:” Insufficient number of observations”

Attached the snap shots of what I did.
Please help.

Thank you

Re: Autocorrelation

Posted: Fri May 25, 2012 10:39 am
by EViews Glenn
Hard to say without seeing the workfile. I do note that you appear to have a three-dimensional panel with 2 observations per block. This may or may not be what you want...

Re: Autocorrelation

Posted: Fri May 25, 2012 11:20 am
by yoyoo
This is the work file.

Are the results change if I use/not use the 3D panel data?

Thank you in advance.

Re: Autocorrelation

Posted: Fri May 25, 2012 4:52 pm
by startz
EViews seems to think the time dimension is your 3rd dimension. You need it to understand that year is the time dimension.

Re: Autocorrelation

Posted: Wed Jun 06, 2012 1:41 pm
by yoyoo
It worked after I deleted the time dimension.

Thank you indeed for your help.