Command "impulse" report
Posted: Wed Jun 01, 2022 6:06 pm
Hi developers,
I am using EViews12. When I used the VAR model, I found the command of "impulse" has some defects:
1. It can only save the confidence intervals that are not calculated from the bootstrap method in the matrix. Hope we can easily save all types of CI in a matrix in the near future.
2. In the official document "Object Reference", the contents about the command "impulse" said that when using Bayesian sampling, we can use the option "seed" to set the random seeds. However, I found we can still use it when using any type of bootstrap. I think it's good, and should be stated explicitly, or users may not fix their random seeds when using other types of bootstrap.
I am using EViews12. When I used the VAR model, I found the command of "impulse" has some defects:
1. It can only save the confidence intervals that are not calculated from the bootstrap method in the matrix. Hope we can easily save all types of CI in a matrix in the near future.
2. In the official document "Object Reference", the contents about the command "impulse" said that when using Bayesian sampling, we can use the option "seed" to set the random seeds. However, I found we can still use it when using any type of bootstrap. I think it's good, and should be stated explicitly, or users may not fix their random seeds when using other types of bootstrap.