There is a known bug that occurs when the steady-state prior is used and the deterministic term is not only a constant (intercept). When subtracting the unconditional mean term from the observed quarterly values, the average within the quarter should be taken. The current implementation uses the final value of the quarter. Evidently, there is no problem when the unconditional mean is constant. Will be fixed in the new version of the package.
There is a known bug that occurs when the steady-state prior is used and the deterministic term is not only a constant (intercept). When subtracting the unconditional mean term from the observed quarterly values, the average within the quarter should be taken. The current implementation uses the final value of the quarter. Evidently, there is no problem when the unconditional mean is constant. Will be fixed in the new version of the package.