Mean values in a log-linearized model

Hi Michel,
I am simulating a large model (2 economies+rest of the world) using Dynare. The model in general works quite well. I log-linearize the model before simulating with Dynare. The steady-state values are all zero by definition and even if I ask Dynare to compute them it gives me zeros for all variables, Fine!
However, when Dynare solves the model it gives means of variables different from 0. I am not simulatig data, I am computing theoretical moments. How can this be explained? and is it really a problem for the rest of the results: irfs and second order moments…
Note: there are price levels in the model.



Hi Nooman,

with second order approximation, the mean is different from deterministic steady state because of Jensen inequality.
Now if you log-linearized by hand before entering the model in Dynare, there shouldn’t be any second order term, even if you requested second order by inadvertance. Maybe there is a typo in your equations and they aren’t linear.



Thanks Michel,
what you just said sound very reasonable. I am going to check again the linerized model.