Otherwise I cannot run the estimation. With an extended model which includes more parameters it seems to be necessary to also exclude more parameters from the estimated params command
I meant: what is the problem/error message you encountered that required you to fix parameters. Typically, that is only valid if there are identification problems.
Dear Professor Pfeifer,
the error message I get is:
Log data density [Laplace approximation] is NaN.
Error using chol […]
If I interpret correctly, the error message indicates that there is an identification problem.
That message in itself can have various reasons. More careful debugging is needed.
I see. Thank you very much for your help, Professor !
Dear Professor，
Can I use a log-linearized model to perform detrending?
In my observation equations, I use expressions like:
data_CC = C - C(-1) + TRENDC;
But after doing this, I noticed that many steady state values are no longer zero. Is this expected behavior? Or am I misunderstanding how to handle detrending in a log-linearized model?
Yes, this is expected behavior. A linear equation may be non-homogenous, i.e., contain a constant.