Three agent NK models

Stochastic simulations
1

I am trying to simulate three-agent NK-DSGE model but dynare keeps failing
Dynare states that
EIGENVALUES:
Modulus Real Imaginary

   2.287e-16        2.287e-16                0
   4.045e-14       -4.045e-14                0
   9.192e-10       -9.192e-10                0
   9.192e-10        9.192e-10                0
       0.606            0.606                0
      0.8794           0.8794                0
       0.948           0.9448          0.07763
       0.948           0.9448         -0.07763
        0.95             0.95                0
        1.01             1.01                0
       1.011            1.011                0
       1.106            1.105          0.05568
       1.106            1.105         -0.05568
       1.215            1.215                0
       39.25            39.25                0
         Inf              Inf                0
         Inf              Inf                0

There are 8 eigenvalue(s) larger than 1 in modulus
for 8 forward-looking variable(s)

The rank condition ISN’T verified!

MODEL_DIAGNOSTICS: No obvious problems with this mod-file were detected.

In my knowledge, BK condition requires equal number of unstable eigenvalues and jump variables and my model have same number of both objects.
I cannot find where the problem comes from. Please help me.

execution.m (1.8 KB)
thrank_nonlin.mod (4.5 KB)

2

No, the rank condition is on top of the number of eigenvalues. It’s about whether you can uniquely determine the contemporaneous variables, which may fail because of a singularity of the relationship between variables. Take the linearized equation system, which typically can be represented as AE_tx_{t+1}=Bx_{t}. If A is singular, there will be an issue. But I have rarely seen models fail like yours. There must be something quite pathological in the relation between your contemporaneous variables.

3

Thanks for kind reply.
Does ‘There must be something quite pathological in the relation between your contemporaneous variables’ mean there might exist some weird or mathematically wrong contemporaneous relationship between the variables I specified?

4

I modified mod file by streamlining some equations and dynare now says

EIGENVALUES:
Modulus Real Imaginary

    3.76e-16        -3.76e-16                0
   2.812e-15       -2.812e-15                0
   6.722e-10       -6.722e-10                0
   6.722e-10        6.722e-10                0
       0.606            0.606                0
      0.8794           0.8794                0
       0.948           0.9448          0.07763
       0.948           0.9448         -0.07763
        0.95             0.95                0
       1.011            1.011                0
       1.106            1.105          0.05568
       1.106            1.105         -0.05568
       1.215            1.215                0
       39.25            39.25                0
   5.269e+18        5.269e+18                0
   1.966e+19        1.966e+19                0

There are 7 eigenvalue(s) larger than 1 in modulus
for 8 forward-looking variable(s)

The rank condition ISN’T verified!

MODEL_DIAGNOSTICS: No obvious problems with this mod-file were detected.

Previous one showed ‘Blanchard & Kahn conditions are not satisfied: indeterminacy due to rank failure’. But now dynare says 'Blanchard & Kahn conditions are not satisfied: indeterminacy’.
Does this message mean previous rank problem is solved?
Is there any way to find what variables are classified as forward-looking variables?

5
  1. The message you are now getting is usually caused by a timing error or wrong parameter setting.
  2. Use model_info to see the variable classification.