Page 1 of 1
contemporaneous terms in VAR
Posted: Wed Jul 25, 2012 10:28 am
by leybert
hello,
I would like to ask how I can add a contemporaneous term to a VAR equation?
Re: contemporaneous terms in VAR
Posted: Wed Jul 25, 2012 2:16 pm
by EViews Esther
Your question seems unclear to me. Can you be more specific?
Re: contemporaneous terms in VAR
Posted: Thu Jul 26, 2012 5:08 am
by leybert
Hi, I have three variable VAR (y1, y2, y3). In the normal case, the output will be how the lags of y1,y2 and y3 affect the current values of y1, y2 and y3. However, i would like to include, for instance the CURRENT y2 and y3 in the equation for y1, the current y1 and y2 in the equation for y3 and the current y1 and y3 in the equation for y2?
Re: contemporaneous terms in VAR
Posted: Thu Jul 26, 2012 6:09 am
by EViews Gareth
Use a System object, not a VAR.
Re: contemporaneous terms in VAR
Posted: Thu Jul 26, 2012 8:59 am
by leybert
can you explain more about exactly how to do that
Re: contemporaneous terms in VAR
Posted: Thu Jul 26, 2012 10:56 am
by EViews Glenn
From your VAR, select Proc/Make System and choose to order your equations by variable or by lag (doesn't really matter except for how you want the coefficients to be ordered). Then edit the System specifications to add in the terms that you want, etc. The Chapter on System estimation will give you more info on the way that systems are specified.
Re: contemporaneous terms in VAR
Posted: Thu Jul 26, 2012 11:12 am
by EViews Glenn
I should also add that if you do this you should think very hard about whether your regressors are still orthogonal to your errors.
Re: contemporaneous terms in VAR
Posted: Thu Jul 26, 2012 12:42 pm
by leybert
Hi,
thanks a lot, but in this case it does not have VAR as an estimation method, any advice on which one i can use. Basically I want to have that current y3 enters the regressions for y1 and y2? Thanks
contemporaneous terms in VAR
Posted: Thu Jul 26, 2012 12:48 pm
by EViews Gareth
VAR estimation is usually just least squares.
Re: contemporaneous terms in VAR
Posted: Thu Jul 26, 2012 2:41 pm
by EViews Glenn
At this point, I'll repeat, for the last time, my recommendation that you think very hard about whether least squares is an appropriate estimator for this specification.