FEVD with calib_smoother

Good morning!

It is good news that in the new Dynare 6 we will be able to do historical decompositions in a calibration, that is, with the calib_smoother command. Please, is there a way that we can also do forecast error variance decomposition in a calibration? With the calib_smoother command?

Best regards,


The forecast error variance decomposition is a theoretical population concept not related to any historical sample. You would need to run stoch_simul on the calibrated model. It will display the FEVD.

Hello Johannes, thank you.

Yes, the FEVD can be obtained with stoch_simu before seeing any data, thank you. However, I was thinking on the conditional forecast error variance decomposition; that is, could I do something like


so I could see the percent contribution of each of the shocks to the forecast errors of a given variable for each of the foreasting horizons, say from 1 to 12 quarters.

Best regards,


You would need to pass the conditional_variance_decomposition option to stoch_simul.

Thank you very much!