Eigen Values

[ebaykal]

Hi all,

I have a mod. file which gives me BK error. Although two of the eigen values reported by dynare =1; they are counted as if they are > 1. Is there a specific meaning as to why these are counted >1 ? Or what does an eigen value which is equal to 1 mean? Normally, I should have as many eigen values >1 as the number of jump variables. However, I am not quite sure about how dynare treat eigen values when they are exactly equal to 1?

I greatly appreciate any help you might offer.
Thanks.

Here is the dynare output :


EIGENVALUES:
Modulus Real Imaginary

9.241e-017 9.241e-017 0
2.888e-016 2.888e-016 0
1.003e-015 -1.003e-015 0
1.219e-015 -1.219e-015 0
6.19e-015 6.19e-015 0
6.311e-015 -6.311e-015 0
1.73e-014 -1.73e-014 0
1.284e-013 1.284e-013 0
1.181e-006 -1.181e-006 0
0.2208 0.2208 0
0.5292 0.5292 0
0.67 0.67 0
0.95 0.95 0
1 1 0
1 1 0
1.01 1.01 0
1.01 1.01 0
1.01 1.01 0
1.508 1.508 0
1.909 1.909 0
4.574 4.574 0
8.554e+005 -8.554e+005 0
Inf -Inf 0
Inf Inf 0
Inf Inf 0
Inf Inf 0
Inf Inf 0
Inf Inf 0
Inf Inf 0


There are 15 eigenvalue(s) larger than 1 in modulus
for 14 forward-looking variable(s)

The rank conditions ISN'T verified!

[StephaneAdjemian]

Hi, a unitary eigenvalue is treated as a "stable eigenvalue" by the command check (even if it means that you have a unit root in your model). In your case, the first unit eigenvalue is counted as a stable while the second is counted as an unstable eigenvalue. This may be due to a round off problem. Can you post your mod file ?

Best,
Stéphane.

[ebaykal]

Thank you very much Stephane.

Here is the mod. file I am working with, "modifiednewapproach_ ramsey".

Is there a way to tell which variable(s) or equation(s) are causing this problem? I tried to modify the model but it did not work.

Thanks again.

Best.

[ebaykal]

Hi there,

I desperately need help with this eigen value problem. I tried to figure out, which variables are associated with eigen values of 1. I remember that I read in dynare manual or forum that the eigen values are given in some order in dr_.eigval. Are they given alphabetized? I was not successful in modifying my model without knowing which variables are causing the problem of more eigen values >1 for less number of jump variables.
I appreciate any suggestion you might offer.
Thanks.