Hi there, I've been working on getting a non-linear Wald test running in Ox. The tricky part is that the restriction is on coefficients of a VAR, as in the following: vC = (ve2'-ve0'-dEta*ve1') *mA* invert(unit(rows(mA))-dEta*mA) - ve3; Here, mA is a coefficient matrix of dimension 4x4. Basically, the above requires use of the Delta Method to compute the Wald Test, hence I would need the Jacobian matrix of numerical derivatives. I realize there is a function for this, but I am unable to get it working - I wonder if it is suitable for the case where the coefficients are on matrix form (rather than vector form)? If anyone knows something about this, it would be a big help! Best regards, Martin