Eigenvalues with Zero Modulus

[reubenpjacob]

Hi All
I am playing with a DSGE model of banking and discovered something strange when I used the check command. Many eigenvalues are zero or very close to zero in modulus. For example,

Modulus Real Imaginary

0 0 0
0 -0 -0
0 -0 -0
0 -0 -0
0 -0 -0
2.956e-18 2.871e-18 -7.009e-19
9.48e-18 9.471e-18 -4.143e-19
6.683e-17 -6.682e-17 8.935e-19
1.116e-16 1.08e-16 2.829e-17
2.254e-16 2.254e-16 5.296e-19
2.992e-16 -2.99e-16 1.143e-17
5.572e-16 5.565e-16 -2.657e-17


I am not sure what the implications are. The Blanchard-Kahn conditions are associated with forward looking variables and eigenvalues above unity. But zero eigenvalues...are they problematic? My guess is that it indicates that many variables do not play a role in the dynamics. Perhaps I am wrong.
Have others experienced the same 'problem'?

cheers
Reuben

[jpfeifer]

In terms of modulus, 0 is just a number like any other and indicates nothing problematic. But you are right that a 0 eigenvalue indicates that the model will have less endogenous propagation than one where the eigenvalue would be ceteris paribus bigger in modulus.