I’m trying to get some forecasts from an estimated model that I have. I tried out both forecast and conditional_forecast, but the unconditional_forecast values that the “conditional_forecast” command gives me are drastically different than the values that forecast command gives. Why would the unconditional forecast from the conditional_forecast command give me a different forecast than the ordinary forecast command?
Thanks a lot!
Could you provide an example to show the problem?
Yes, sure. The task I am trying to do is the following. I have a neo-keynesian model with 1 lag that is calibrated to some set of data that goes until 2013 q3 in a separate step. I want to take 2013q3 as the initial point and see what the model forecasts for the next 40 quarters. I tried using the forecast command, which gave me some odd results, so then I decided to see what conditional_forecast gives me as an unconditional forecast as just a basic check.
Here’s the relevant code snippet:
After the model block and the shocks block
options_.order = 1;
options_.jacobian_flag = 1;
options_.nonlin = 1;
//Setting equal to 2013 q3 data
then for the last line I either put
conditional_forecast(parameter_set=calibration, controlled_varexo=(eR),periods=40);[/code] or [code]forecast(periods=40);
The values that I get from forecasts.uncond.Mean in conditional_forecasts.mat are quite a bit different from oo_.forecast.mean in Dynare_edo_results.mat , which seems counter intuitive to me. Does anyone have an explanation for that?
I had a quick look at it and it seems there is a problem with the unconditional_forecast values that the “conditional_forecast” command gives due the initival values not being recognized. If you are after unconditional ones, please use the forecast command. I will keep you posted when we have fixed or resolve the issue.
thanks a lot for the response. Does this mean that the conditional forecast values are also not working? The output I get for that is also a straight line…