I have the issue with the steady state - I have been working for quite some time now on my model and I haven’t been able to resolve this. The post is related to this one but I have been working on solving the complete model with two sectors, rather than simplified one I posted earlier. I solved for the steady state separately but when I input those values into initval block, Dynare reports that the steady state has NaNs or Inf. This is possibly due to some negative values I obtained in the steady state file, I tried changing the values of some parameters but was not able to fix it. Would really appreciate any suggestion.
Warning: Some of the parameters have no value (i_bar) when using steady. If these parameters are not initialized in a
steadystate file or a steady_state_model-block, Dynare may not be able to solve the model. Note that simul,
perfect_foresight_setup, and perfect_foresight_solver do not automatically call the steady state file.
I tried to solve the model from the original paper I built the extension to, in order to figure out the issue with my model. However, even for that paper, where the authors solved the model in Matlab, if I use their steady state values as initval block in Dynare I face the same problem with steady state. Would you perhaps have any suggestion why this happens?
I would appreciate any help as I really can’t figure this out. Attached are the relevant codes from their replication package and the .mod file I wrote.
Are these correct? The first implies in steady state:
1=(1-rhos)*(1-rhoOmega)+n*(1-rho);
which is violated with your initial values. AT the same time, the bracketing is very different from (A.79) in your linked PDF. At this stage, I would recommend setting up the model with \LaTeX definitions and checking all equations. Mistakes like the one above may cause the steady state not to exist.
Thank you very much once again! I am actually not sure, I noticed there are few equations in their simplemodelSS_Targets_Final.m file which are not consistent with the linked PDF file. I tried to solve that model in Dynare thinking that would help me figure out the issue with my own model, but it’s not the case.
That’s the tricky part with replications: you don’t know whether the printed equations and the computer code are consistent, and if not, which one is actually correct. But such inconsistencies will typically derail any attempt at finding a steady state.
True, but I am not sure how to figure out the issue with my model because there I can check for inconsistencies and it still does not work unfortunately.
Yes, it’s the file MinWageModel2.m from the post above. Here are the updated files (I changed low sector wage to be also determined endogenously rather than exogenously but it did not help):
Are you even sure that there is a unique steady state for your model? Your model features price levels, which are usually indeterminate in NK models. Only relative prices and inflation are uniquely determined.
Yes, there should be. But, maybe that’s the mistake that I wrote all the equations with w/P for instance, instead of substituting it with real terms (w_r).
Yes, that will definitely be a problem for numerical steady state finding. The steady state will not be unique, because e.g. only the ratio w/P is unique, but not the individual components.
I actually defined a new variable PH_P to equal P_H/P and then initialized those. However, in the optimal price setting equation I have this term varpsi which I can’t set the initial value for:
You need to fully replace such terms. With respect to \Psi, you need to alter the recursive representation to eliminate the single price level. Usually, you can just multiply it through on both sides.
