2SLS procedure in EViews
Posted: Sun Mar 02, 2014 5:55 am
Hey,
despite 4 semesters of Finance studies, the curriculum somehow contained almost nothing on simultaneous equation models and absolutely nothing on panel data and 2SLS.
Now I have to use all those techniques and feel like I have no idea what I'm doing. The general equations in my Econometrics books don't seem to correspond exactly to my equations and the explanations are not practical enough.
I have the following relationship I would like to investigate and read that simultaneous equations w/ 2SLS is the best method to do that.
1) ROA = a0 + a1*LEV + a2*FAM + a3*FAM*LEV + a4*1/2*LEV^2 + a5*SIZE + e
2) LEV = b0 + b1*ROA + a2*FAM + a3*FAM*ROA + a4*1/2*ROA^2 + b5*SIZE + u
So ROA and LEV are my two endogenous variables, possibly influencing each other. All others are assumed to be exogenous.
1. Is the model just identified, overidentified or underidentified?
2. In either of the three cases, what implications does that have for estimation requirements and inference from the results?
3. How do I get the reduced form of the equations?
4. How do I enter this in EViews (what goes in the estimation specification box, what in the instruments box?)
5. If EViews only gives me coefficients for equation 1), how will I know the coefficients for equation 2) ? (see this link: there is only one result box, showing the coeffs. for one equation. http://staffweb.hkbu.edu.hk/billhung/ec ... app09.html )
Any help is much appreciated. Thanks!
gerrard
Using EViews 7
despite 4 semesters of Finance studies, the curriculum somehow contained almost nothing on simultaneous equation models and absolutely nothing on panel data and 2SLS.
Now I have to use all those techniques and feel like I have no idea what I'm doing. The general equations in my Econometrics books don't seem to correspond exactly to my equations and the explanations are not practical enough.
I have the following relationship I would like to investigate and read that simultaneous equations w/ 2SLS is the best method to do that.
1) ROA = a0 + a1*LEV + a2*FAM + a3*FAM*LEV + a4*1/2*LEV^2 + a5*SIZE + e
2) LEV = b0 + b1*ROA + a2*FAM + a3*FAM*ROA + a4*1/2*ROA^2 + b5*SIZE + u
So ROA and LEV are my two endogenous variables, possibly influencing each other. All others are assumed to be exogenous.
1. Is the model just identified, overidentified or underidentified?
2. In either of the three cases, what implications does that have for estimation requirements and inference from the results?
3. How do I get the reduced form of the equations?
4. How do I enter this in EViews (what goes in the estimation specification box, what in the instruments box?)
5. If EViews only gives me coefficients for equation 1), how will I know the coefficients for equation 2) ? (see this link: there is only one result box, showing the coeffs. for one equation. http://staffweb.hkbu.edu.hk/billhung/ec ... app09.html )
Any help is much appreciated. Thanks!
gerrard
Using EViews 7