TSLS and serial equation
Posted: Fri Jul 23, 2010 5:24 am
Hi to all,
I am running a simple OLS with two endogenous variable in it (both dummy), and I am struggling with the TSLS as Eviews (6) define it:
my original equation is (simplified):
Y=A+B+C+D+E+F
BUT
A=C+G
B=C+G
I am running a TSLS on Eviews with the following equation:
first equation (first part of the windows)
Y c A B D E F
second equation (instrument list)
c C G D E F
Is it right? why the R-squared is always negative?
I was thinking at a second solution:
I first regress singularly A and B and then I add the residuals (or fitted) as variable in the equation of Y. That is a simple OLS of:
Y=resA+resB+D+E+F
is it correct?
I am running a simple OLS with two endogenous variable in it (both dummy), and I am struggling with the TSLS as Eviews (6) define it:
my original equation is (simplified):
Y=A+B+C+D+E+F
BUT
A=C+G
B=C+G
I am running a TSLS on Eviews with the following equation:
first equation (first part of the windows)
Y c A B D E F
second equation (instrument list)
c C G D E F
Is it right? why the R-squared is always negative?
I was thinking at a second solution:
I first regress singularly A and B and then I add the residuals (or fitted) as variable in the equation of Y. That is a simple OLS of:
Y=resA+resB+D+E+F
is it correct?