Steady state detection


Hey there, I have a problem to find the steady state of “Financial business cycles” Iacoviello (2015). Before I estimate the shock-parameters, I want to stoch_simul the extended model. To make things “easier” I only take into account 2 shocks and set the others to 1 (they are multiplied) and their equations are in log. For the missing parameter-values that are needed I take the posterior mean of Iacoviellos estimation-results. I calculate the initial values from Iacoviello’s steadystate.m file and wrote my own .m file as an alternative.
Unfortunately no steady state can be found.
Is this approach correct?

I would appreciate every hint :slight_smile:

Find attached my .mod file and Iacoviellos equations. Will upload my steadystate.m later if possible (I’m a new user and can only upload 2 files)

extmodel.mod (7.1 KB)

FBC_equations.pdf (132.3 KB)


Also find attached my steadystate.m file

extmodel_steadystate.m (3.6 KB)


The issue is not that no steady state is found, but rather that the steady_state.m-file you provide does not return the correct analytical steady state matching the entered model equations.


Thank you for your reply jpfeifer! How do you see that the steady_state.m file doesn’t fit the model (except that no steady state could by found?)
I adapted the .m file from the reproduction file set from the original paper. But I noticed that the equations in the .mod file for reproduction of the model differ from the equations given in the paper / technical appendix.
Do you have an idea why that’s the case?

Thank you for your help

  1. Dynare gives you the residual conditional on the steady state file. They are not 0. You need to check what you altered and whether the original files before you changed them run as expected.
  2. No, I don’t know. What exactly is different?


Thank You jpeifer.
For example: In the published version of the paper in the setting of the extended model equations B.26 and B.27 contain the expression (1-rho_e)* m_n* A_me* lambda_e.
In the technical appendix which is in the folder “reproduction files” on Iacoviellos website, the same equations are listed without (1-rho_e) and in addition the .mod file for reproduction doesn’t contain the shock A_me. In my understanding that’s important for the propagation of the shock troughout the model… Besides that the weights in the production-function (1-mu) and mu are written the other way round.