I have a simple question about rolling windows. I have quarterly data, but I’m looking to advance it forward by year. I know I can use nobs and first_obs = [integer1:integer2] to do a rolling window estimation, but this moves it quarter by quarter. I would like it to move by year (or every 4 quarters), for example 2000 - 2010, 2001 - 2011, 2002 - 2012, etc…
I tried using first_obs = [1:4:15] as that seemed intuitive, but it doesn’t work.
What exactly do you want to do with the results from the rolling windows?
What Dynare does when you specify
nobs as more than a scalar is to loop over
dynare_estimation. You could simply run your own loop here.
Basically see if the parameters change over different periods of time.
I think running my own loop is the best option as this point, I was just checking if there was an easier way to do it that was already implemented in Dynare.
Thank you so much for the response! Big fan of your github page.