Hi Everyone,
After estimation, I got two smoothed shocks that are similar. Is there any possible reason to cause such a problem?
These shocks reflect the movements in the underlying data. As such it’s not completely surprising some similarities appear. The model will try to ascribe big movements in the data to more smaller shocks than a single big one. That being said, big correlations between shocks can happen in sample, but may also indicate model misspecification.
Hi Dr. Pfeifer,
Thanks for replying. If the model is misspecified, how could we check which parts of the model cause this problem?
That often involves a lot of trial and error, e.g. testing which features help. Sometimes, you can add measurement error to the estimation and see from the size of the measurement error where the model has trouble explaining the data.