I have several ARDL models; they are well specified. But, there are correlations among residuals of equations. So I need to estimate a SUR system. But, I cannot solve the SUR system using systemfit package in R. Regarding on reference file for the package, I suppose having equations with a different number of observations could be the problem. Because I am trying to solve 11 different ARDL models and all models includes different lag structure, it is not the case for me to work with balanced equations. Do you know a package in R which can work with equations with different observation lenght?