Re: [Gretl-users] Converting VAR forecast coefficent

Mirko, The answer depends on what transformations you're using (if any) in your model. If X is in millions, then for example if X_t+1 = a + b*X_t, then a, b and c have the usual (regression) interpretation as marginal effects. A $1million increase in X today would raise your forecast for next period by $b million. If your model is in logs or growth rates, then you'd have to apply the inverse transformation to get back to $. Note that things can get tricky if your transformation is nonlinear though. For example, if your model is x_t+1 = log(X) = a +b*x_t, then your forecast of X_t+1 is exp(a+b*x_t + 0.5s^2) where s is the sample standard deviation. HTH, PS -----Original Message----- From: gretl-users-bounces(a)lists.wfu.edu [mailto:gretl-users-bounces@lists.wfu.edu] On Behalf Of mastro2387(a)tiscali.it Sent: Wednesday, March 13, 2013 12:22 PM To: gretl-users(a)lists.wfu.edu Subject: [Gretl-users] Converting VAR forecast coefficent Dear all, I have a question. Let us suppose I do a VAR to forecast some values of my variable X that is measured in Millions of Dollars. After having done all the process, how can I 'translate' the coefficients of the forecast so as to have the right value in Millions of Dollars? I hope to have explained the issue in an understandable way. Kind Regards, Mirko Mastrolia Invita i tuoi amici e Tiscali ti premia! Il consiglio di un amico vale più di uno spot in TV. Per ogni nuovo abbonato 30 € di premio per te e per lui! Un amico al mese e parli e navighi sempre gratis: http://freelosophy.tiscali.it/ _______________________________________________ Gretl-users mailing list Gretl-users(a)lists.wfu.edu http://lists.wfu.edu/mailman/listinfo/gretl-users