Gambacorta, Signoretti (2014) replication

Hi all,

I am trying to replicate Gambacorta and Signoretti model (2014) - a shorter version of Gerali et al (2010), by using the steady ratios referred in the paper and try to solve for the steady state and put them in the initval block in Dynare.

However, i get an error like this: The steady state contains NaN or Inf.

Is it a proper method to use those ratios mentioned in the paper (i.e C/Y=0.9, I/Y=0.11, c_e/C=0.05 etc) and general rules like I=δ*Κ, labour L=1/3 (0.33) and then solve for the steady state?

Am i completely missing something? Is there any issue with the equations? Would a matlab file solving for the steady state be easier, although i am completely unfamiliar with that?

Attached you may find the model.GS_2014.mod (3.5 KB)

Kind regards

Recheck all equations.

A_e=rho_A_e*A_e(-1)+e_A_e;  % exogenous process
m_e=rho_m_e*(m_e(-1))+e_m_e;
mk_y=rho_mk_y*(mk_y(-1))+e_mk_y;

imply that all these variables have steady state 0. But that is for example inconsistent with

y_e=A_e*k_e^ksi*l_p^(1-ksi);   % production function

which should be

y_e=exp(A_e)*k_e^ksi*l_p^(1-ksi);   % production function

Thank you @jpfeifer. I modify the exogenous process equations to

A_e=(1-rho_A_e)1+rho_A_eA_e(-1)+e_A_e;
m_e=(1-rho_m_e)m_e_ss+rho_m_e(m_e(-1))+e_m_e;
mk_y=(1-rho_m_e)1+rho_mk_y(mk_y(-1))+e_mk_e,
and use the exp() in the production function, but again i receive the same message.

You mean that i need to check all the equations (not just the exogenous process)?

Morever, i am not sure why i should use the exp() in the production function.

Finally, the approach of using the steady state ratios and solving the system is appropriate?

Kind regards

  1. You need to either set the steady state to 1 by adding a constant term or put exp() in the production function (in the production function you will have exp(0)=1 in steady state then), but not both.
  2. Yes, you should check all equations. Mistakes there often prevent finding a steady state
  3. That approach is correct but not generally applicable. Usually you need to have the steady state computed to know those ratios. But if you are replicating a paper, then if the paper is correct and you start from their correct ratios it should work.

Thank you very much @jpfeifer. I will try to follow your advices and hopefully find a solution.