model with recursive preferences will not solve

[phdstudent99]

Hello,
I am trying to solve a model with recursive preferences similar to the one in Colacito and Croce (JoF 2013).
My issue is that the model cannot be solved for values of risk aversion greater than 2.
The error message I obtain depends on the value of risk aversion AND on the value I enter manually for the qz_zero_threshold.

For example, the model is solved when risk aversion (SIGMA) is 2 and the irfs seem to make sense.
When SIGMA =3 and qz_zero_threshold = 1e-10, I get the following error:
One of the eigenvalues is close to 0/0 (the absolute value of numerator and
denominator is smaller than 1e-10!
If you believe that the model has a unique solution you can try to reduce
the value of qz_zero_threshold.

When SIGMA = 3 and qz_zero_threshold = 1e-20, I get the following error:
There are 1 eigenvalue(s) larger than 1 in modulus
for 2 forward-looking variable(s)
The rank condition ISN'T verified!
dynare:k_order_perturbation: Caught Kord exception: The model is not Blanchard-Kahn stable

When SIGMA = 4 and qz_zero_threshold = 1e-20, I get the following error:
There are 2 eigenvalue(s) larger than 1 in modulus
for 2 forward-looking variable(s)
The rank condition is verified.
dynare:k_order_perturbation: Caught Kord exception: NaN or Inf asserted in first order derivatives in FirstOrder::solve
Error using print_info (line 68)
k_order_pert was unable to compute the solution

Does this mean there is a problem with my model or with my code?
My mod file is attached and I am using Dynare 4.4.3 and Matlab 2013a.

Thank you.

[jpfeifer]

There seems to be a more fundamental collinearity/singularity problem in your model. You can see this when running

Code: Select all

model_diagnostics

As detecting singularity as opposed to near-singularity is a numerical challenge, it can pop up in strange ways. That is what you encounter.