Smoothed variables and shocks

I’m a novice in DSGE modelling and a bit confused regarding smoothed variables and shocks. Questions that pop up include when and (when not) for what purpose to use them etc? Smoothed var/shocks seems rather standard in the literature and I want to deepen my knowledge. However I have a hard time finding good guides, papers etc that dig deeper in the subject.

I’m solely looking for some good references to deepen my knowledge.

Also, when estimating a model in Dynare, the smoothed variables and shocks is calculated by default without any specific setting in the estimation command?


The idea of the Kalman smoother is to provide best estimates of the unobserved data and shocks, given all the observable data. They can be used to answer various typical questions in the literature. For example, which type of shock has most likely cause the Great Recession. You need to trigger them with the smoother-option of estimation.

Thank you for your reply.

I have seen estimations without the smoother option followed by a stoch_simul with the estimated parameters and the calib_smoother in another model file, is this a common proceedure?

Do you have any additional references which I can read upon?


It depends on what you are trying to do. Estimation will provide Bayesian objects like the posterior distribution of the smoothed objects. Sometimes people prefer to have the smoothed objects at the posterior mean. That’s when you use the calib_smoother.