Page 1 of 1

SEM solving

Posted: Mon Oct 10, 2016 12:10 pm
by aam
Hello Gareth and all
Im using EViews 7 and trying to solve a model with 10 stochastic eqs and 17 identities...ive estimated the 10 eqs by 2sls but when solving the model i get
2002Q4 NA generated for Y_SA - Log of non positive number
Solve terminated - Log of non positive number
i really dont know whats going on... (i can post the workfile if necessary)
Thanks a lot
aam

Re: SEM solving

Posted: Mon Oct 10, 2016 12:57 pm
by EViews Gareth
You will need to provide the workfile.

Re: SEM solving

Posted: Mon Oct 10, 2016 1:47 pm
by aam
Hello! ive attached the workfile........when i solved the model without gelr/(py_sa/100) in eq 13, it works fine......i really dont know
thanks in advance!!
aam

Re: SEM solving

Posted: Mon Oct 10, 2016 1:58 pm
by EViews Gareth
There isn't much to say.

Equation 4 specifies that Y_SA is a function of DLOG(X), i.e. Y_SA = log(X)-log(X(-1)).

If X goes negative, this will result in an error since you can't take the log of a negative number. The model converges to a position where X goes negative.

Re: SEM solving

Posted: Mon Oct 10, 2016 3:25 pm
by aam
i had everything in millions of pesos from 2004 and the variable i added for the simulation not.........so that was making x (effective demand) negative.............sorry for troubling you and thanks a lot!!!
agustin