Dear Gretl Team,

I would like to give 2 suggestions that I think could make VAR estimations easier (at least for the GUI users :).

In the VAR model window, under "Tests -> Autocorrelation", it would be great if we could have a more "complete" test result. Please, take a look at this example: using "australia.gdt" data and estimating a VAR(1) model with the variables "le", "lpus", and "lpau" (I'd attached a Gretl session file) we get the following output:

Equation 1:
Ljung-Box Q' = 3.56914 with p-value = P(Chi-square(4) > 3.56914) = 0.467

Equation 2:
Ljung-Box Q' = 6.00246 with p-value = P(Chi-square(4) > 6.00246) = 0.199

Equation 3:
Ljung-Box Q' = 33.5337 with p-value = P(Chi-square(4) > 33.5337) = 9.29e-07

I think it would be better if Gretl could present not only the lag number 4, but all the lags requested by the user (just like it does in an OLS estimation - "Graphs -> Residual correlogram"):

LAG      ACF          PACF         Q-stat. [p-value]

    1  -0.0141       -0.0141          0.0157  [0.900]
    2  -0.0333       -0.0335          0.1047  [0.949]
    3   0.1832        0.1825          2.8302  [0.419]
    4  -0.0947       -0.0949          3.5691  [0.467]
    5   0.0827        0.0987          4.1402  [0.529]
    6   0.0553        0.0139          4.3989  [0.623]
    7  -0.1578       -0.1228          6.5371  [0.479]
    8   0.0775        0.0465          7.0605  [0.530]
    9   0.0920        0.0878          7.8091  [0.553]
   10  -0.1368       -0.0960          9.4905  [0.486]
   11   0.0357       -0.0034          9.6066  [0.566]
   12  -0.0637       -0.0736          9.9830  [0.617]


The second suggestion is related to the graphs options (in the VAR model window, "Graphs"). Why don't we put the residual correlogram option? ;-)

Thanks in advance for your attention,
--
Henrique C. de Andrade
Doutorando em Economia Aplicada
Universidade Federal do Rio Grande do Sul
www.ufrgs.br/ppge