NKM estimated values are strange

Estimation ML/Bayesian estimation
1

Hi,

I am running estimations on a NK model with non-zero steady state inflation, ascari-ropele (2007) typy. The model is loglinearised and I use Euro Area data for now (please see the attached files). My problem is that the estimated values are out of what I expected and also what the literature says. The parameter sigma (inverse of intertemporal elasticity) as well as the parameter for habit formation (h) are too low, eta (inverse of the elasticity of labour supply) is too high. I also have some issues with the TR parameter phi_y. Can someone help me please? (I attach the code and data files)

Or would it be preferable that instead of structural parameters, I estimate composed parameters of the phillips curve, xi_f, xi_b, kappa …?

And my last question: why is dynare running the same estimation three times? (please see the attached .log saved as text file)
EA_str.log.txt (24.2 KB)
EA.xls (61 KB)
EA_str.mod (3.69 KB)

2

I cannot do the debugging, but the 3 runs of estimation are explained by

which triggers recursive estimation for samples of 86 to 90 observations. I am also puzzled by you not using demeaned output for estimation

3

the data for output are given as the output growth, a seasonally adjusted growth rate of GDP in constant prices. thats where the observation equation y_obs = y-y(-1) comes from. Is it wrong?

4

But you did not demean output growth in the data and do not allow for a constant in your observation equation.

5

thank you for your reply. Originally I thought that the constant in observation equation cancels out if I use output growth. Now I did demean the output growth (using HP filter), however the results did not change much. The estimated values are pretty much as before. Could it be caused by something else?
Thanks a lot in advance!