EViews.com

I have a fixed-effect model
log(cash) = f(cf, nwk, log(size), rdev, political, svtraded)
Where
Political is a dummy that equal 1 if company is indentifed as politically connected and zero otherwise.
Svtraded is the ratio of the stock value traded to GDP that varies over time but constant for all company.
When I run this model, Eviews announced "Near singular matrix"
Can you help me?
I attached my data here.

Is the dummy constant for each firm?

No, Political equals 1 for several firm and equal 0 otherwise, and Svtraded varies over time but constant for all company. For expamle, in 2008, Svtraded equals 20, in 2009, Svtraded equals 30, but in 2008 it equals 20 for all company

If you run the model without fixed effects, does it estimate?

If political is always the same for a given firm, you're in the dummy variable trap.

But if I run without fixed-effect, I have to create many dummy variables. Can u give me another way. Here is the literature that I'm studying. Thanks for your help.

Perhaps you want to read up on the dummy variable trap. You seem to want to do something that is impossible. It has nothing to do with EViews per se.