Forecast in a calibrated model

Stochastic simulations
1

Dear all,

I would like to make conditional forecasts in a calibrated model, using the calib_smoother command, then, after it, obtain the initial conditions as the last obs filtered state, and then forecast.

To this end, I used the calib_smoother command with the options parameter_set = calibration and filtered_vars. Then, I used smoother2hist(outfile = initial_conditions_for_forecasting), and histval_file(datafile = initial_conditions_for_forecasting) after it.

Finally, I called the conditional_forecast with the parameter_set = calibration.

The forecast is indeed executed but with the starting value as the steady-state instead of the last obs.

The smoother2histval command successfully creates the .mat file initial_conditions_for_forecasting.mat, but for some reason the histval_file command is not setting the correct initial conditions.

The mod-file and data are attached

Thanks in advance !

df.xlsx (35.7 KB)
US_model.mod (6.2 KB)

2

For anyone out there that may have the same issue, I’ve observed that you don’t need to specify priors before estimation command if you activate the smoother option.

After that, if you use conditional forecast command with parameters_set = prior_mean he will do the proper forecast with the correct initial conditions.

Even though, the issue above appears to be a bug.

3

I can confirm that this is a bug. We are working on it. Glad to know you found a workaround.