The root forecast mean squared errors after rolling window

Estimation ML/Bayesian estimation
1

Hello everybody

I have a question that does DYNARE 4.5
Dynare 4.5 can run the rolling window, but after that can DYNARE compute the Root Mean Squared Errors (RMSEs) ? or we have to compute it manually?
If Dynare can not do that, then we have to compute the RMSEs manually, so where is the predicted value stored for each window?
in Dynare Manual, it introuduces that the predicted value is stored at oo_recursive. I used it, but I found nothing, only thing below appears
’oo_recursive_
oo_recursive_ =
Columns 1 through 9
[1x1 struct] [1x1 struct] [1x1 struct] [1x1 struct] [1x1 struct] [1x1 struct] [1x1 struct] [1x1 struct] [1x1 struct]
Columns 10 through 16
[1x1 struct] [1x1 struct] [1x1 struct] [1x1 struct] [1x1 struct] [1x1 struct] [1x1 struct]’

there are 16 columns since I rund 16 windows

can someone help me in this case?

Thank you so much for that

2

You have to compute it manually. oo_recursive is a structure storing the oo_ results structure from each run in the respective entry. So oo_recursive{1,1}.PointForecast should contain the point forecast.

3

Dear Prof. @jpfeifer

I got it. In particular, I used it as command

oo_recursive_{1,16}.MeanForecast.Mean.y_obs

it implies that I would see the mean forecast of variable y_obs at the 16 Window in 4-horizontal forecast.
Dynare reports
ans =
0.0018
0.0017
0.0015
0.0014

My understanding is correct? If not, please correct me
Thank you so much for that

4

Yes, exactly. The 16 should refer to the window starting with observation 16.

5

Thank you so much for that prof. @jpfeifer