Eigenvalues vs. fwd-looking variables comp


I’m trying to replicate Akinci & Queralto (2019) (code attached).

Dynare is telling me I have 39 eigenvalues with absolute value above 1 for 38 forward-looking variables.

In contrast to previous posts pointing to similar issues, by counting the number of variables that enter with a lead I think the 38 number is right. Does this necessarily imply there is an error with the calibration / steady state? (which would in turn affect the eigenvalues)

The model_diagnostics function says there are no obvious problems with the model.


Akinci_Queralto_medium_forum.mod (31.4 KB)

Usually this is a sign that you made a timing mistake somewhere in your model.

Yes, thanks! Managed to find that last mistake. All running smoothly now.

Thanks again.

Hi Silvio,
I have your same problem. Do you remember how you solved?
Many thanks

@Dynaresauro There is no general fix. I can be mistakes in the equations like wrong timings or signs. Or wrong parameter values.