I wanna take log of differenced data in GMM.
If I represent variable like log(), and click differences in step 3.
Will the software difference the log transformation?
It makes no sense to do that.
It there a way to difference the data first, then transform to log?
Thanks
Search found 3 matches
- Fri Oct 15, 2010 10:00 am
- Forum: Econometric Discussions
- Topic: GMM differencing
- Replies: 0
- Views: 1739
- Mon Oct 04, 2010 1:02 pm
- Forum: Econometric Discussions
- Topic: test the lag length of panel data using GMM
- Replies: 1
- Views: 4363
Re: test the lag length of panel data using GMM
I was wondering should I do a restricted and unrestricted GMM for each lag length
I am not sure about the restriction
like for lag length 2
should it be A(-1) A(-2) B(-1) B(-2)
or A(-1).......A(-38) B(-1).....B(-38)
I am not sure about the restriction
like for lag length 2
should it be A(-1) A(-2) B(-1) B(-2)
or A(-1).......A(-38) B(-1).....B(-38)
- Sat Oct 02, 2010 12:43 pm
- Forum: Econometric Discussions
- Topic: test the lag length of panel data using GMM
- Replies: 1
- Views: 4363
test the lag length of panel data using GMM
My objective is test bivariant granger causality for panel data according to Holtz-Eakin's paper: if the increase in the sum of squared residual is "large" then the optimal lag length can be determined. However, there is no "big jump" in my output, so I can not draw any conclusio...