So far, the only way I know is to export data to GNU R and plot FEVD there. This approach, however, has its disadvantages, as one has to estimate the whole system in R first and R's estimates often differ significantly from those obtained with gretl.
This is more worrying. The same model should produce the same estimates both in gretl and R. Could you provide us with an example?
First of all, thank you so much for the script - works perfectly fine for me!
As far as the issue with different estimates is concerned, this is probably all my fault (I'm quite new to gretl and completely new to R), but still there's a slight chance that something might be wrong.
In gretl I do:
vecm 2 1 X Y Z --ct
The above produces exactly the same output as STATA's:
vec x y z, trend(trend) /vecrank and lag order given in specification window/
and slightly different form JMulTi's estimates.
However, when I use R's (urca):
xyz.vecm <-
ca.jo(gretldata[, c("X", "Y", "Z")], type = "trace", ecdet = "trend", K = 2, spec = "transitory") /for all options in 'type' and 'spec'/
and transform obtained VECM into VAR representation:
vec2var.xyz <- vec2var(xyz.vecm, r = 1)
the system I get has quite different parameters from those estimeted in gretl, STATA or even JMulTi (obviously I do not compare VAR to VECM - there are diffreneces in VAR representations).
Still, this might be some kind of my mistake, so if you see anything wrong in the code I presented above, please let me know.
Best,
Greg