Initial values in estimation

Stochastic simulations
1

Dear Prof. Johannes Pfeifer,

I am just repeating the estimation code of Iacoviello(2015, RED). I only replace the data of Iacoviello’s, and the code does not work. It reported that:

(minus) the hessian matrix at the “mode” is not positive definite!
=> posterior variance of the estimated parameters are not positive.
You should try to change the initial values of the parameters using
the estimated_params_init block, or use another optimization routine.

I don’t know how to set the initial values of the parameters, could you help me to check it and tell me how to solve it?

Bfbc2.mod (13.9 KB) fbc2_steadystate.m (4.5 KB) MYDATA.mat (3.1 KB) est,
Xiang

2

Using mode_compute=5 improves the situation. But then ZRHO_AP runs to the upper bound.

3

Thank you professor!
I have changed the code to mode_compute=5, but it reports the same error as before.

POSTERIOR KERNEL OPTIMIZATION PROBLEM!
(minus) the hessian matrix at the “mode” is not positive definite!
=> posterior variance of the estimated parameters are not positive.
You should try to change the initial values of the parameters using
the estimated_params_init block, or use another optimization routine.

Log data density [Laplace approximation] is NaN.

I don’t know how to solve it. Could you please help me?

Best,
Xiang

4

As I said before

You need to find out why that is the case. Maybe adjust the prior.