Re: [Gretl-users] FEVD plot in gretl and GNU R
2011/3/28 Sven Schreiber <svetosch(a)gmx.net>
The usual suspects are: double-check the sample which is used, the lag
order (seems to match here though), and the precise specification of
deterministic terms (both restricted and unrestricted). Don't know the
details of urca to tell you right now.
-sven
The sample and lag order are correct for sure. My choice for deterministic
terms is "unrestricted constant and unrestricted trend". This might be the
issue, as in gretl and STATA you can specify this particular option (aka
'case 5'), while in JMulTi and R the specification for deterministic terms
seems to be more general and I'm not quite sure whether ecdet="trend" in
urca is an exact equivalent of "--ct" in gretl... Is it?
Attachments:
- attachment.html (text/html — 1.1 KB)