Equating results for simple RBC model


I am new to Dynare and after reading through the Dynare User Guide I am experimenting with different methods of model specification. I have estimated the same simple RBC model using both the “model” and “model(linear)” specifications, however I am getting different steady state conditions for the two models. I’m posting to ask: “why are the proposed steady states different?”.

The model is a simple RBC model with an aggregate productivity shock. The pdf file “Summarized_RBC_MODEL.pdf” contains a breakdown of the model specification, steady state conditions, calibration and log-linearized equations. An extended explanation of the model is contained in the file “RBC_Model.pdf”. The two mod files “RBC2.mod” and “RBC_LINEARIZED2.mod” contain the “model” and “model(linear)” specifications respectively.

Thank you for your time,
RBC_MODEL.pdf (235 KB)
RBC2.mod (2.01 KB)
Summarized_RBC_MODEL.pdf (151 KB)

With a linearized model, the actual variables are percentage deviations from steady state, while the steady state values of the original variables become parameters as in

Yss*Y = Css*C + Iss*I;
As we are talking about different concepts here, the steady state values will of course not coincide. With the original nonlinear model, Dynare is going to compute the steady state numerically. With the linearized model, it is your job to provide correct values for the parameters (see also my “Guide to Observation Equations” for details)