Blanchard & Kahn conditions for less forward-looking variable(s)

Stochastic simulations
1

Dear, professor jpfeifer,

I always got the error “There are 11 eigenvalue(s) larger than 1 in modulus for 10 forward-looking variable(s). The rank condition ISN’T verified!”

I simplifed the model as best as I can to figure out why, I checked the timing, FOCs, parameters values, however, the simplifed version still got the same error.

I find two complex eigenvalues, is it a problem?

A very simple version of my DSGE model only run when I set the parameter of interest sticking to inflation \rho_\pi below to 1. It can not run if I set to 1.5, which is common in the paper.

I update the files and hope you can run it and give me some suggestions on why these happen.

Thank you so much!

finalcode2026.mod (13.9 KB)

finalcode2026_steady.m (3.5 KB)

finalcode2026calib.m (3.6 KB)

2

This typically signifies a problem in the timing setup. Given your sectoral structural, that is complicated to debug.

3

Thank you for your quick response, professor, you are always kind and helpful.

The simple version of my model has removed all the complicated features and is almost the same as Iacoviello (2005, 2020). However, I always got the error above. Can I ask you three more question?

First, I find two complex eigenvalues, is it someting need to worry?

Second, In Iacoviello (2005, 2020), Household and firm choose the housing demand, however, the housing supply is normalized to 1 each period, is it a problem?

Third, how can I debug the simple version of my DSGE model to make it run, can you give me some suggestion?

Thank you so much! professor.

4
  1. No, complex generalized eigenvalues are nothing to worry about.
  2. No, having housing in fixed supply does not pose a conceptual problem as shown in those papers.
  3. Given that none of your model versions work, it is usually best to build your model up from a working simple model like Iacoviello 2005 instead of removing features from a huge model that does not work.
5

Thank you for your quick response, professor. I think the simple version is quite standard and almost the same as Iacoviello 2005, however, rank condition ISN’T verified! I hope to run a simple version first then add other features of my DSGE model but stucked here. Anyway, I will follow your suggetions and try a few more days. I really appreciate your help. Best wishes!