Weird mode check plot, why?

Estimation ML/Bayesian estimation
1

Dear professor

I am implementing a Bayesian estimation, and I get weird mode check plots.
mode check plots.pdf (7.2 KB)
mode check plots2.pdf (9.1 KB)

What is the reason? Here is my estimate code and data.
est.mod (17.8 KB)
data1204.mat (7.1 KB)

Thank you very much

2

Is there maybe a steady state file missing?

3

Sorry , I forget
est_steadystate.m (5.1 KB)

4

Please use the attached files with Dynare 6.2. When I remove the prior bounds you set, it runs:
est.mod (17.8 KB)
est_steadystate.m (4.7 KB)

5

Thank you very much Professor jpfeifer for your reply.

I ran the code according to your suggestion and finally I got the following result.
mode check plots1.pdf (8.4 KB)
mode check plots2.pdf (5.9 KB)

First, the result of pgidgdp_p is weird, I don’t know whether there is a problem?
Second, pgidgdp_p, that is, the response coefficient of monetary policy to output, is 0.7930, which is so large that I think this estimate is not reliable. May I ask why?
Third, other mode check plots are flat, such as the standard deviation of all shocks. What does this tell us? Whether there is a problem.
I have a lot of questions, thank you for your patience.

6
  1. The red dots indicate Blanchard-Kahn issues. I would not worry too much about them given the scaling of the x-axis. The problems appear to be rather far away from the mode.
  2. That is hard to tell. But something in your data (or model) pushes the Taylor rule towards low inflation and high output feedback. I don’t know whether that is expected for your sample.
  3. The plots are not really flat. Look at the scaling of the y-axis.
7

Hi all. I wonder what mode_check plots with negative values on the y-axis mean. I got ones too but I thought those numbers should be non-negative because the y-axis represents density…

Thank you in advanced!

8

It’s the log density. See Bayesian estimation question - #5 by jpfeifer

1 Like