Am 19.04.2013 13:03, schrieb artur tarassow: > Hi all, > > I estimated a VAR model using the standard var-command and want to run a > function called varmodtest afterwards. The function simply exploits the > "modtest" command and merges different specification test results into a > matrix which is passed back. > > <hansl> > function matrix varmodtest (scalar order "Lag length") > #Serial correlation > modtest order --autocorr --quiet > matrix OUTSC = $pvalue' | $test' > > return OUTSC > end function > > var 2 LRM IBO > matrix A = varmodtest(4) > <\hansl> > > But I obtain the error: > "Can't do this: no model has been estimated yet > Data error > *** error in function varmodtest" > > How can the function grab information from the previous VAR estimation? Not the solution, but perhaps a viable workaround: include the estimation of the model itself in the function. -sven _______________________________________________ Gretl-users mailing list Gretl-users(a)lists.wfu.edu http://lists.wfu.edu/mailman/listinfo/gretl-users

Am 19.04.2013 13:26, schrieb artur tarassow: > Hi Sven, > yes of course that would be the standard "solution", but makes things, > maybe unnecessarily, more complicated. > But I would like to know just out of curiosity whether there is a more > direct way. > possibly another workaround: don't use a function, but put the modtest stuff into another script which you invoke via 'run' (untested). -sven _______________________________________________ Gretl-users mailing list Gretl-users(a)lists.wfu.edu http://lists.wfu.edu/mailman/listinfo/gretl-users