Hello everybody,
for computing higherorder approximations and being able to use the error function in the model equations I wanted to try Dynare++. Playing aroung with the simple RBC model in its plain vanilla type (CRRA utility function with consumption and labor, CobbDouglas production function, TFP shock, no frictions) I was not able to get the results of Dynare++ and Dynare in accordance. The model is approximated to the 2ed order. The following counterintuitive results occured:

Dynare++ produces means for consumption, investment, capital and production that are below their deterministic steady states. Dynare does the opposite. As it is not allowed to upload .mat files I just give the mean values for the capital stock:
Dynare: 11.4907; Dynare++ 11.3885; deterministic steady state: 11.4297. 
Adding a second shock variable (Variable “b”) in Dynare++ that does not occur in any other model equation changes the results of Dynare++ for all variables, e.g. mean capital stock: 11.5538.
I do not unterstand these results.
The model files for both Dynare (PlainVanilla.mod) and Dynare++ (PlainVanilla++.mod, PlainVanillaB++.mod) are enclosed. I suppose that I make a simple general mistake in Dynare++ that I just don’t see. It would be very helpful if you could help me out.
Many thanks,
Peter
PlainVanilla.mod (2.36 KB)
PlainVanillaB++.mod (817 Bytes)
Plainvanilla++.mod (788 Bytes)