I managed to run a mod file despite model_diagnostic says it has 3 colinear relationships, IRFs and moments are calculated successfully.
In other words, when I run order=2 simulation, Rank condition is verified.

Therefore it is a bit strange that the model_diagnostic complains I have colinearity. It also uses the word ‘static’ in the following phrase:

[quote]model_diagnostic: the Jacobian of the static model is singular
there is 3 colinear relationships between the variables and the equations[/quote]


Am I correct in saying that the model_diagnostic is only looking at colinearity for order=1 simulation? Thus I may still be able to run order=2 simulations on such a mod file

As the name col-linearity implies, this is a property at first, i.e. linear order. And it refers to the static model, i.e. in steady state. Essentially, your model does not uniquely determine the deterministic steady state. For most models, this is a bug indicating a deeper issue like a problem with Walras Law. If this is the case, you cannot run order=2.
The same message will be present if the model has a unit root as in this case, there are many possible steady states. In open economy models, this regularly happens when the model is not properly closed. See Schmitt-Grohé and Uribe 2003, “Closing Small open Economy models”. In this case, the foreign net assets are non-stationary.
If you want to preserve this feature and the collinearity is thus intended, you can run order=2.