Am 20.01.2014 16:28, schrieb Allin Cottrell:
Octave and JMulTi are potential problems. (BTW, is there still a case for
handling JMulTi data?).
Well JMulti isn't developed any further, but you might also simply call
it "mature software". I guess nobody knows who is using it and how
often. My personal needs have been mostly covered by gretl since the big
leap forward in (restricted) cointegration analysis and also the SVAR
package (although I confess I haven't had to use the latter for serious
work yet). Plus the cointegration stability test package that Andreas
wrote on my request.
OTOH there is still special functionality in Jmulti:
* Smooth transition models (STAR) -- anybody knows where to find those
in other software (just out of curiosity, no real need yet)
* unit root tests with trend breaks (probably in CATS)
* bivariate GARCH, IIRC (I don't think this is covered by gretl packages
like gig, no? -- should be in recent Eviews versions I think, and I
guess R?)
Wasn't it the case that Jmulti had actually adopted some Gauss file
format? If so, one could actually ask the question why gretl hasn't
advertised its import capability of Gauss files, but not sure if it's true.
-sven