> Hi all, > > I'm trying to use gretl's vecm command. The following statement : > > vecm 2 1 x1 x2 > > gives me the following output : > > Cointegrating vector : > x1(-1) b1 > x2(-1) b2 > > Equation 1: d_x1 > const theta > d_x1_1 g1 > d_x2_1 g2 > EC1 alpha > > Given the above output, I'd expect the predicted value d_x1_hat for d_x1 to be : > d_x1_hat = theta + g1 d_x1_1 + g2 d_x2_1 + alpha ( b1 x1(-1) + b2 x2(-2) ) It should be "alpha (b1 x1(-1) + b2 x(-1))", but I suppose it was just a typo. Other than that, I can't see anything obviously wrong.

> However, applying the above formula yields crazy results. Surely, I > must be overlooking something. Any hint ? What do you mean exactly by "crazy results"? Can you be more specific? Can you reproduce your problem with, eg, the "denmark" data set supplied with gretl?