Re: [Gretl-users] impulse-response functions from VECM in Gretl
John C Frain
3 Aranleigh Park
Rathfarnham
Dublin 14
Ireland
www.tcd.ie/Economics/staff/frainj/home.html
mailto:frainj@tcd.ie
mailto:frainj@gmail.com
On 19 August 2017 at 09:43, lasses skola <lassesskola(a)gmail.com> wrote:
Thanks for an instructive reply! Yes, the non-normality is due to a
few
outliers. If I am to use this in a paper, what problems, caveats or other
circumstances do I need to adress/mention?
How you deal with outliers depends on the causes and nature of the
outliers. Are
they caused by different methods of calculation of the
aggregates involved? Are they caused by some change of policy or by a
technical innovation? You Are they best represented by an ordinary dummy or
a step dummy or some other process. Juselius in her book "The cointegrated
Var" explains the use of various types of dummies in her examples. It was
fashionable at one time to look at recursive (rolling) estimates of these
models to examine the effect of the outliers. If you are submitting to a
journal it is likely that a referee will have read this book. Your question
is more about your data and statistical practice rather than Gretl.
_______________________________________________
Gretl-users mailing list
Gretl-users(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users
Attachments:
- attachment.html (text/html — 2.4 KB)