Page 1 of 1

Maximum number of equations in a system

Posted: Fri Oct 09, 2015 9:36 am
by leah
Is there a limit to the number of equations you can add to a system? I get a " near singular matrix" error when I try to estimate a system using SUR and more than 100 equations.

Re: Maximum number of equations in a system

Posted: Fri Oct 09, 2015 9:40 am
by EViews Gareth
There isn't a limit per se, but if you add too many variables, you'll eventually hit a maximum number of elements in a matrix for EViews to handle. This will be a function of workfile size, number of equations and number of variables. But, for example, I can create a SUR with 200 equations without hitting that limit fairly easily:

Code: Select all

create u 1000 !n =200 system s coef(!n ) alpha coef(!n ) beta for !i=1 to !n series y!i=nrnd series x!i=nrnd s.append y!i=alpha(!i) + beta(!i)*x!i next s.sur
Your singularity problem is more likely to be caused by a true singularity.

Re: Maximum number of equations in a system

Posted: Fri Oct 09, 2015 9:48 am
by leah
Yup, that was it -- my sample had exactly 100 observations. Thanks for your help!