Strange behaviour in unobserved variables

I am a beginner in using dynare for estimating DSGE model. I developed a pretty standard (log-linerarized) DSGE model and I estimated it using dynare. All the posteriors density, MDD, and the other diagnostic graphs seem to be ok. However, if I plot the unobserved variables, they are not centered around their mean. So I suspect I am doing something wrong. How can I figure out what is the problem? Any hints are welcome.


If all your variables in the model are mean 0, but the smoothed variables/shocks are not mean 0, then you did a mistake in your model somewhere. In particular, the steady state of the entered model variables (including the observables) cannot be 0.