Page 1 of 1

near singular matrix error message

Posted: Fri Jan 04, 2013 11:19 am
by xinchao
Hi,

I am new to Eviews and try to learn how to use it for my research topic. When I run panel regression (fixed effects), I have encountered a problem which is the "near singular matrix" error message.

My equation in Eview is as follows: PNAV C SIZE LEV LIQ CUR

Everything was ok until I added one dummy variable, TYPE, which is equal to 1 if it's a non-REIT company & 0 if it's a REIT.

However, when I run with unstructured/undated data, there is no problem when I add this dummy variable. The error message only appears when I run panel regression. I don't know what the problem is and how to solve it. Is there anyone encountered this problem before? Can you please help me with this?

I also attached my data here. Thanks a lot.
SES panel data.xls
(161.5 KiB) Downloaded 366 times

Re: near singular matrix error message

Posted: Fri Jan 04, 2013 2:50 pm
by EViews Gareth
Probably perfect colinearity between the fixed effects and the dummy.

Re: near singular matrix error message

Posted: Fri Jan 04, 2013 7:05 pm
by xinchao
so that means I can't include this variable in my regression, right?

Thanks a lot.

Re: near singular matrix error message

Posted: Fri Jan 04, 2013 10:21 pm
by EViews Gareth
Right