Hello I have about consumption preference

[yoyoNDHU]

Hello!
I have a question!
I know maybe it not a good question.., because it's not a question for code.
My question is in my model my consumption is

C_t= [(1-alpha_t)^(1/eta)C_ht^((eta-1)/eta) + (alpha_t)^(1/eta)C_ft^((eta-1)/eta)]^(eta/(eta-1))

I want set alpha_t be a AR(1) shock, but someone ask me
if you set it be a shock if alpha >1 your model will get wrong~~
It's will happen in dynare? If yes, how can I set a regine for the shock~~

Thanks for everyone

[jpfeifer]

I don't completely get the question, but Dynare computes a smooth Taylor approximation to your FOCs. A constraint of the type you are imagining would require an occasionally binding constraint. This cannot be done in Dynare. Maybe it could be done using the extended path method.

[yoyoNDHU]

Really really thanks for your help!!

I want ask two more question!

1) I know order=1 is using in log linear form

but...I'm so stupid use in nonlinear form

It will get a big error??

2) Because of using nonliear form

I remember dynare will take log by itself

then I set AR(1) process to A=A(-1)^rho*exp(ea)

finally my model is success to get result but I afraid all the result are wrong..

thanks for your help!

[yoyoNDHU]

Does someone can give me the answer or some suggest?

Thanks for everyone!

[jpfeifer]

1) Nobody knows. The approximation error depends on the strength of the non-linearity and the shock size. You may be lucky or not.
2) Dynare won't automatically take the log. You have to do this manually be using a variable substitution with exp(). Dynare does the linearization for you.