After adding the “identification” commend, mod can not run

Dear professor jpfeifer,

My mod file run well without the “identification” commend. (mode_compute = 4, Dynare 4.6.1 version), parameters’ starting values equal prior means.

However, after adding the “identification” commend, all parameters are identified but hessian matrix at the “mode” is not positive definite and the mod file can not run any more.

I only add the “identification” commend, other details are identical, therefore, i really don’t know why this problem occurs and I am really confused.

I am wandering if the mod file without the “identification” commend gives me completely wrong model results.

I provide the mod files with and without “identification” commend, and hope for your reply which always help me a lot, thank you.

without identification comend.mod (19.3 KB)
with identification comend.mod (19.3 KB)
unemployment2022.xlsx (17.7 KB)

In Dynare 5.0, both versions run, but the one after identification finds a better mode. So it may have to do with local maxima. Running identification first will usually result in a different random number seed that typically will not matter but seems to do so here.

Thank you, professor jpfeifer. You are always kind and your replies are always useful and means a lot for me. thank you.