About steady-state of non-linear DSGE model

Hello. I’m running non-linear DSGE model and I got some question about the steady-state.
At first, I tried to get a steady-state with ‘steady’ command but it said the steady state cannot be computed.
So I solve the steady-state analytically as in the code I attach below, then the steady-states are calculated, although I’m not sure they are right numbers.

Then my questions are:

  1. Can I trust my derivation of steady-state and proceed the simulation?
  2. If I can trust it, why I failed to get a steady-state with the ‘steady’ command? Was it the problem with the initial value?
  3. If I cannot trust it, then are my model equations has some mistakes?

Thank you very much for your help!

nonlinear_model4.mod (8.0 KB)

See your resid numbers in Dynare output results.
Resids should be zero for all equations. Altough this is not a straightforward method.

When you use steady command in Dynare you can see steady state or mean values for the endogenous variables in Dynare output results.

You can calculate steady state values with different algorithms in Dynare, although you have calculated them analytically.

Thank you for your reply! I got further questions.

With the analytical steady-state equation method, I checked residuals and it was all zero.
Then it means I can trust my analytic result?

With the ‘steady’ command, I tried with different algorithms but I failed with everything.
Is it the problem with the initial value or problem with my model equations?

I can not say your analytical steady state values are correct completely because I did not see your DSGE model and your calculations.

Now, when you run your model in Dynare is there any error in output results ?

What I meant is that whether I can proceed to simulation part when I get steady-state with ‘steady_state_model’ block, but not with ‘steady’ command.

With ‘steady_state_model’ block including analytic steady state values, I got no errors.
All residuals are zero and model diagnostics said “No obvious problems with this mod-file were detected.”

Thank you.

That means your analytical steady state indeed solves the entered model equations. Of course there could by still mistakes in the equations, but it’s a first sensible sanity check that is not available when using a numerical steady state finder.

I see. I really appreciate the help of both of you!

Thank you.