Dynare noob here.
I have two questions about the interpretation of smoothed variables after an estimation (or even with calib_smoother). Suppose I have a model with output, capital, hours worked, and consumption. I feed in data for output, consumption, and hours worked, and afterwords Dynare generates smoothed capital. Is smoothed capital interpreted as deviations from its steady state? Should its plot look stationary? The smoothed variable for capital from running fs2000.mod for instance is not centered at zero and fails a Dickey Fuller test (i.e. cannot reject unit root null hypothesis).
Second, is there a mechanism by which I can generate confidence intervals for the smoothed variable? I’m thinking in terms of something like Laubach and Williams where they use a basic NK-ish type model with Kalman filters to infer the natural rate of interest and provide confidence intervals for it as well.