Page 1 of 1
HOW I APPLY CAUSALITY TEST IN PANEL DATA
Posted: Mon Sep 23, 2013 2:16 pm
by ANDROO1988
HI
i need apply causality test in panel data. my equation is : y= a0+a1 X + a2 Z + ... + E
that x , z are major variable in my model. now, how i can test causality between (Y,X,Z).? I have eviews 7
Re: HOW I APPLY CAUSALITY TEST IN PANEL DATA
Posted: Mon Sep 23, 2013 3:16 pm
by EViews Gareth
EViews 7 does not have panel causality testing built in. EViews 8 does.
Re: HOW I APPLY CAUSALITY TEST IN PANEL DATA
Posted: Tue Sep 24, 2013 12:45 am
by ANDROO1988
thanks. how i apply this test in eviews8? can you tell me it step by step? please.
Re: HOW I APPLY CAUSALITY TEST IN PANEL DATA
Posted: Tue Sep 24, 2013 6:38 am
by EViews Gareth
Open a group, click on View->Granger Causality.
Re: HOW I APPLY CAUSALITY TEST IN PANEL DATA
Posted: Tue Sep 24, 2013 7:17 am
by ANDROO1988
ok. thanks. i test it.
Re: HOW I APPLY CAUSALITY TEST IN PANEL DATA
Posted: Thu Sep 26, 2013 6:09 am
by James71
Dear Gareth,
I can´t perform a Granger causality panel test with Eviews 8. I do not why. Could you help me?
As usual when working with panel data, i have created a pool object (in fact I've taken a Eviews 7 workfile). Then I make a group with the two panel variables of interest (string ?), and then: View-Granger causality... and I get what I got with Eviews 7: individual granger tests for all the potential combinations of individual variables. What is the mistake I'm making?
Best
Re: HOW I APPLY CAUSALITY TEST IN PANEL DATA
Posted: Thu Sep 26, 2013 6:46 am
by EViews Gareth
You need a panel workfile, not a pool.
Re: HOW I APPLY CAUSALITY TEST IN PANEL DATA
Posted: Thu Sep 26, 2013 7:18 am
by James71
Many thanks for your extremely quick answer :)
You need a panel workfile, not a pool.
Re: HOW I APPLY CAUSALITY TEST IN PANEL DATA
Posted: Thu Feb 27, 2014 5:58 am
by stijn545
I'm also doing Granger causality on panel data. I was wondering, if I have data for variables A and B in 2010 and 2013, is the following then correct. First, I apply a Granger causality (using zero lags) on A-2010 and B-2013, and then i do the same for B-2010 and A-2013. The only significant F-statistic out of the 4 resulting ones is that A-2013 does not cause B-2010, while all the others are not significant. Then I'm having a hard time coming to conclusions too; what useful information can I draw from a statement which only tells me that A-2013 does not cause B-2010?