Is the algorithm for deterministic simulation robust to models without forward-looking variables? I think it is not, but would like to double check with someone.
If I deterministically simulate (use “simul” command) a model without a forward-looking term, the code crashes because a matrix “iy_” is of an order (2 x n). If I “open sim1”, which is the function that contains the simulation procedure, I see on lines 11 and 14 that “iy_” must be of an order (3 x n). I presume that the matrix “iy_” summarises the characteristics of endogenous variables. If I introduce a dummy in to my model “0endo_var(+1)" (and sometimes also "0endo_var(-2)”) to “fool” Dynare, everything is running, and I am getting plausible results (at least I think so 
)
I attach an example that I have been working on. It is a basic form of the Solow-Swan growth model.
Thank you for any thoughts
Martin
solow.mod (619 Bytes)