Dear all,

I have set up a DSGE model including housing allowing for substitution between rental and owner occupied housing.

I have solved for the steady state analitically following the basic ideas from the technical appendix of Iacoviello & Neri 2010. However, due to aforementioned substitution, I initially solve for three variables numerically in my steady state file using fsolve.

My model does not seem to be able to find a steady state, but I am not sure, where I am wrong. Do you have any suggestions on how to get further from here? I have attached my used files and am usign Matlab2021 and Dynare 4.6.3.

Thanks in advance!

ssfun.m (720 Bytes)

MODELFNG_steadystate.m (4.7 KB)

MODELFNG.mod (5.5 KB)

The steady state file has to supply the correct steady state, if that is not the case it wonâ€™t work. Have you tried using them as initial values?

Thanks for your reply!

I have not tried to set them as initial values in the .mod file directly. The steady is solved using different ratios. Do I â€śjustâ€ť insert these ratios as defined parameters in the .mod file and then apply the solution as initial values, or what is the way to do it?

Thanks!

With the initval version you have to give an initial value for each endogenous variable in your model. If you miss one, Dynare assumes zero as initial value, which could lead to issues.

But yes, overall you just set

initval;

C = ;

C1 = ;

.

.

.

PSI = ;

end;

and hope for the best Look in the documentation and the forum for some more tips on steady state solvers etc.

Okay, I understand now, thanks! But this would imply simply just â€śguessingâ€ť some initial values, and replace the calculated steady state in the steadystate file that I am using, right? Because I would think that chances of guessing a system of 39 values correctly are tiny or have I misunderstood something?

Kind of. The thing I meant was to put in the calculated numbers as initial values. Dynare then starts from there and is hopefully able to find a steady state.

The thing is that your calculated steady state does not work, otherwise the steady state file would not complain. But probably you are close, which means starting from there is a good idea.

So what you should try is using the numbers that you had in the steady state file as initial values and see what happens.

Great, Iâ€™ll give it a try - thanks!

Unfortunately, it didnâ€™t work, but thanks for the suggestion!

Still same error message - how would one proceed from here?

Can you provide the most recent version of your file please?

Of course - here you go. I have also attached the SS file used to find the initvalues used in the .mod file.

MODELFNG_initval.mod (6.1 KB)

MODELFNG_steadystate.m (4.7 KB)

ssfun.m (720 Bytes)

Looking at the residuals, based on your initial values, I would say that you are still too far from the steady state that is why it wonâ€™t work. Maybe go over your calculations again and try to find mistakes.

You said that you could solve the most part (all except three variables) analytically, maybe there is a mistake somewhere, because then finding the remaining three numerically should not be the big issue.

1 Like

You have a couple of equations with rather big residuals. Usually that is a sign of either a mistake in the steady state computations or these equations.