multiple non linear equation to solve in helper/steady state

[georam11]

Dear Prof Pfiefer,

Thank you for your reply.

Also may I ask if I use the .m file to find the initial values using fsolve function? and then manually feed the initial values in mod file to be used in Dynare? Should that ideally work?

Since my file contains multiple non linear equations to be solved in fsolve, I cannot find an example of a steady state file which tries to fsolve multiple equations. Also example 3 in Dynare uses a helper file, is that doable as well? But again I cannot find a helper file which solves multiple non linear equations in dynare.

Many Thanks
Ammu
Re: Query on .m file
Sent: Wed Mar 22, 2017 11:42 am
by jpfeifer

Dear Ammu,
the authors initially used Dynare++, not Dynare. The former does not take steady_state-files. If you plan on using Dynare instead, you should use a steady_state-file.
Nevertheless, you finding complex parameters is a conceptual problem. There must be some issue here that you need to solve.
Query on .m file
Sent: Wed Mar 22, 2017 11:09 am
by georam11

Hello Professor Pfeifer,

Hope you are well.

I contacted the authors for the dynare code for KS model and found that they create a Mod file by running a .m file. I have incorporated the frictions on labor and have refurbished the model (See the attached .m file).

However, I encounter few problems.

I solve for the steady state values in matlab but find that the solutions are complex roots.
These steady state values are saved as parameters and used as initial values in the mod file.

The authors have created this .m file in 2011. Do you think I should alternatively try to build a .steady state file/helper file to solve for steady state?

I have rechecked the model many times and find no error in the equations.

Looking forward to your reply

Many Thanks
Ammu

[jpfeifer]

1. Providing initial values for Dynare's steady state should work, but is only the second-best option.
2. Using a steady_state-file will work, while the "helper"-function will have trouble with the multivariate approach. For the steady_state-file, you should be able to call fsolve on your multivariate system the way you already do it.