Non-zero residuals with solver

Stochastic simulations
1

Dear All,

I am trying to fully calibrate my model before estimation. I use solver for steady state. Code runs and solves but I get some non-zero residuals and non-zero steady state values which shouldn’t be there due to loglinearization. I checked the non-zero residual equations but I wasn’t able to see the mistakes.

I appreciate for any idea.

Thank you

Code_constant_calibration.m (1.5 KB) Code_constant.mod (5.4 KB)

2

Why are the variables in exp() if you linearized?

3

Professor Pfeifer,

I think I used wrong term.

I simply scaled equilibrium system by steady state such as;

\hat{X_t} \equiv \frac{X_t}{X}

Thank you

4

Then why should the steady state value be 0?

5

Because I used log deviation from steady state. That is why I used exp() in the equations. For example, I have Css as non zero, but C needs to be zero in the steady state.

6

You should try to get the nonlinear model to work. There is no point in doing the substitution you do.

7

Professor Pfeifer,

Thank you for your suggestion.

8

Professor Pfeifer,

I fixed the equations. Molde runs without any problem now. Apparently I mistakenly used non-linear and linear equations together. I worked on non-linear model and got it work.

Thank you for your help