There are more eigenvalues > 1 in modulus

I’m trying to replicate a NK model with wage stickiness as in the Costa’s book, but now without log-linearizing it. I converted infinite sums as in the optimal wages & prive level equations into recursives. However, this gives me error “There are 6 eigenvalue(s) larger than 1 in modulus for 5 forward-looking variable(s).”

I searched for similar error and have indication that it might be due to “timing convention” issue. However I have no clue as to which variable to be shifted one-period in which equation and direction?

Enclosed the mod file.
costa01.mod (1.6 KB)

Are they using the same parameters values ?

You will notice that if you increase phi or theta, you don’t have this problem anymore.

If you remove the qz_threshold option you specified the model diagnostics will tell you there are some collinearity issues with equations 7, 8, 9, 10. These are lines 33-36. Substituting the factor prices (W and R) in to your equation for what I think is MC=marginal cost of course gets you exactly the production function, hence the singularity: you have entered exactly the same equation twice.

Your initval block also has two expressions for T and P_star, lines 54/64 and 55/65, respectively. I’m not sure that this will be a problem, but it might be worth fixing.