Page 1 of 1
How to incorporate loss aversion function
Posted:
Sat Aug 30, 2014 10:56 pm
by jmlacson
Hi,
I'm trying to incorporate a loss aversion component into my asset pricing model. This requires a function that, I believe, has an if-then statement inside the function. I've searched the user guide and examples but have found none that could help guide me. Would this require an "external" if-then MATLAB code?
Can anybody point me to the right direction?
Thanks, guys.
Re: How to incorporate loss aversion function
Posted:
Sun Aug 31, 2014 8:15 am
by jpfeifer
Please elaborate. For what is the if-condition? Does it depend on parameters or variables. If it depends on variables and introduces a non-differentiability, then you cannot use it for stochastic models. If it is for parameters, then you can use the macro-processor's if-clause.
Re: How to incorporate loss aversion function
Posted:
Sun Aug 31, 2014 11:08 am
by jmlacson
Hi, thanks for the quick reply, Mr. JP.
My consumption CAPM for the household optimization is
beta*((C^(1-gamma)/(1-gamma))+(theta*(((1-N)^1-chi)/(1-chi))+(beta*(v(X(+1),Z))))
My problem is the third component: (beta*(v(X(+1),Z)))
the variables are X(+1) and Z.
X(+1) = R(+1)-R_f
If X(+1) > 0, then (v(X(+1),Z))
If X(+1) > 0, then (lambda*(v(X(+1),Z))
Based on your question, it appears my issue fall under variable dependent and non-differentiable.
Does that sound right, Mr. JP?
Thanks and regards,
Jomar
Re: How to incorporate loss aversion function
Posted:
Sun Aug 31, 2014 12:40 pm
by jpfeifer
That sounds like a Karush-Kuhn-Tucker problem/occasionally binding constraint. You cannot use this type of constraint in a stochastic model in Dynare as the kink at 0 introduces a non-differentiability. Essentially, you need to assume that you are always in one of the two cases to use Dynare.
Re: How to incorporate loss aversion function
Posted:
Sun Aug 31, 2014 2:02 pm
by jmlacson
Thanks for your inputs, Mr. JP.