Warnings in estimation results

Estimation ML/Bayesian estimation
1

Dear professor:
There are lots of warnings when I add observed variables and estimation block in my model. The Warning is “Matrix is close to singular or badly scaled. Results may be inaccurate”.
There are no problems when I simulate the results of shock.

code1112A.mod (8.3 KB)
However, warnings appear when I add observed variables and estimation block in my model.

code1112A1.mod (9.2 KB)
sm1115.xlsx (11.3 KB)
Though there are lots of warnings, the MODEL_DIAGNOSTICS command show “No obvious problems with this mod-file were detected”.
I wonder whether there are some problems in my model.
Thanks!

2

It seems your model is running into issues when one tries to change theta_s. The mode_check-plots look rather strange. What is the interpretation of that parameter? Is it expected to introduce such complex dynamics?

3

Thanks for your reply！
My model is used for steel industry. The parameter theta_s is elasticity of substitution between the capital-labor composite and iron ore. I want to estimate this parameter by
bayesian estimation. Could you give me some advice?

4

Could you give me some advice about this problem? Thanks a lot!

5

Have you checked whether that parameter is well-identified? Most often, people calibrate those parameters.

6

Thanks. I try to check it.

7

Dear professor,
I use the identification command and get this figure. I think this parameter is well-identified.
Can this result be accepted? What should I do next？
Thanks in advance!

image
image865×649 22.3 KB

8

Is that the same code as Bayesian estimation process has many warnings?

9

No, they are different. I add a shock in this code, then the paremeter can be identified. However, the mode_check -plots look rather strange.

10

If this parameter can satisfy identification, mode check, and diagnostic，how can I deal with these warnings?

11

The warnings are something you may ignore. I am more worried by the strange mode_check-plots.

12

Thanks a lot! Your reply is very helpful to me. I’ll try to change the prior distribution and observed data.