Estimated natural wage markup

Hi, all

i´m trying to estimate the model proposed in Galí, Smets and Wouters “Unemployment in a estimated New Keynesian model”, 2012 (GSW)
using Spanish data. the model includes the same observable series that in Smets and Wouters, 2007 including
unemployment rate instead of employment rate. I follow the Smets and Wouters’ code developed by Michel Juillard, adding
some variations. For simplicity, the model doesn´t include the rule for smoothed consumption nor the exogenous labor supply.

The program works correctly and values of estimated parameters are mostly close to the same values obtained with other models like (Casares, Moreno, Vázquez, 2014). However, some estimated results doesn´t make any sense. i.e, estimated natural wage mark-up is close to zero in every period.

Equations included in GSW which allow us to define the labor market are the following:

Wage setting
w = …+ w4*(marw-marnw) ( only the term in which appears unemployment is written.)

Wage markup:
marw = w - (z + eps_x + sigma_ll) = sigma_lunem

Natural wage markup:
marnw = 100 * eps_w = sigma_l* unem_star;

Measurement equation:
unemobs = unem + costunem; (costunem is the Steady-State unemployment rate)

I can´t find the way to write in Dynare correctly how to obtain reasonable values for the natural wage markup. I think the measurement equation related to observable unemployment is not suitable defined according this definition of unemployment in GSW, but i couldn´t find how to define in a different way.


As i had no found the way to make it works, i tried to calibrate the model as in Galí, Smets and Wouters, 2012. “Unemployment in a estimated New Keynesian model”
but again i have problems:

“Impossible to find the steady state. Either the model
doesn’t have a steady state, there are an infinity of
steady states, or the guess values are too far from the

The message is concerning the equations defining consumption with flexible prices either sticky prices:

STEADY: numerical initial values or parameters incompatible with the following equations
7 20

Do you have some idea about what is wrong in this definition of the model??

GSW_calibra.mod (7.45 KB)

Your model is supposed to be linear, but this is not the case in equations 7 and 20. For example in equation 20 you have

cf = ((lambda/tau)/((1+(lambda/tau))*cf(-1)))  

If you check the bracketing, you will see that you divide by cf(-1), making the equation clearly nonlinear.

Thanks a lot.

Hi Pedro
Did you sove the problem with your code? I have a problem also to write the code to replicate the paper Unemployment in a estimated New Keynesian model. If you don’t mind would you share it?