# Do we need to compute steady state when doing estimation

Hi!
when doing bayesian estimation,we just calibrate few parameters.so if we do not log-linearization the model, we do not have to compute the steady state right?

The model is solved by computing a first order Taylor approximation around the steady state. So the steady state is always needed. The thing is that you may not need to provide an analytical steady state to Dynare as the program may be able to compute it numerically. But usually it is still recommended to compute the steady state by hand.

1 Like