Em 25 de março de 2010 Allin Cottrell <allincottrell@gmail.com> escreveu:
Henrique wrote:

> I'm estimating a VAR model and I'm getting a strange Gretl behaviour. When I
> try to choose the optimal VAR lag using the GUI (Model  -> Time series ->



> VAR lag selection) Gretl crashes (I attached the Mac OS X Crash Report). 

Can you give some more details? I can't replicate this problem so far.
What I did, in the gretl GUI (on Linux and OS X):

1) Open data9-12 (a monthly dataset)
2) From the Add menu, add log differences of the first 4 variables
3) Select Models/Time Series/VAR lag selection
4) Select the 4 log-differences as Endogenous, and accept all



5) Click OK
6) Reasonable-looking results are shown
Dear Allin, I realized what is going wrong. Take a look at the following script:

<script>
open data9-12
ldiff pcecars pop pcdpy cpinew
var 24 ld_pcecars ld_pop ld_pcdpy ld_cpinew --lagselect

/*
Teste para verificação do erro na seleção de defasagens
do VAR via GUI.
*/
</script>

In that case everything is working fine. But if I change the script subtracting
the last line (*/):

<script>
open data9-12
ldiff pcecars pop pcdpy cpinew
var 24 ld_pcecars ld_pop ld_pcdpy ld_cpinew --lagselect

/*
Teste para verificação do erro na seleção de defasagens
do VAR via GUI.
</script>

Now the problem arises. I think it would be important to introduce an error
message instead of a crash, won't you? ;)

Um abraço,
Henrique C. de Andrade
Doutorando em Economia Aplicada
Universidade Federal do Rio Grande do Sul
www.ufrgs.br/ppge