You should include enough lags to get rid of autocorrelation, so that shouldn't be an issue. Chapter 25 of the Gretl User's Guide gives good examples on VARs. The script on var --lagselect (page 199) will be of use. There is also a nice user interface: Models>Time Series>Vector Autoregressions. Check is out with the User's guide close by. On Thu, Apr 4, 2013 at 12:43 PM, aymen kaabi <kaabiayman(a)gmail.com&gt; wrote:

---------- Forwarded message ---------- From: aymen kaabi <kaabiayman(a)gmail.com&gt; Date: 2013/4/4 Subject: multivariate autoregression To: gretl-users(a)lists.wfu.edu good evening. i have project to do about multivariate autoregression using cross and auto correlation . i find the procedure GRETL_VAR <http://gretl.sourceforge.net/API/old/gretl/gretl-libgretl.html#GRETL-VAR:... * gretl_VAR <http://gretl.sourceforge.net/API/old/gretl/gretl-var.html#gretl-VAR> (*int order*, *int *list*, *const double **Z*, *const DATAINFO <http://gretl.sourceforge.net/API/old/gretl/gretl-libgretl.html#DATAINFO:C... *pdinfo*, *gretlopt <http://gretl.sourceforge.net/API/old/gretl/gretl-libgretl.html#gretlopt> opt*, *PRN <http://gretl.sourceforge.net/API/old/gretl/gretl-PRN.html#PRN:CAPS> *prn*, *int *errp*); if this procedure do what i need. please explain to how can i use it. thanks fol all