panel TSLS instrument selection

[bobchowz]

Hi everyone,
Now I am working on my paper, there is some problem with the Eviews specification input and instrument input.
My structure equations are:
1. FDIportion=c+ln(PCRGDP)+dummy1+dummy2+dummy3
2. ln(PCRGDP)=c+FDIportion+humancap+popgrowth+ln(infrastructure)

Q1:basically I want to avoid the bidirectional impact between FDIportion and PCRGDP, that's why I introduce three dummies. Does that mean these three dummies are all FDIportion's IV?
Q2: I ran TSLS for each equation, for EQ1, I put FDIportion c log(PCRGDP) d1 d2 d3 in specification, and c d1 d2 d3 humancap popgrowth log(infrastructure) in instruments; for EQ2, I put log(PCRGDP) c FDIportion humancap popgrowth log(infrastructure) in specification, and d1 d2 d3 humancap popgrowth log(infrastructure) in instruments. PLZ help me figure out what's wrong with it.

Your help is appreciate!

[startz]

Why do you think anything is wrong?

[bobchowz]

Why do you think anything is wrong?

thank you sir.

I quote from my professor:
"You still don't do 2SLS correctly. First off, there is no first stage. There is only a second stage, the first stage being estimated in the background. More importantly, the instruments MUST be the EXOGENOUS or PRE-DETERMINED variables in the structural equation plus perhaps other exogenous or pre-determined variables. You've got to decide which of the explanatory variables in the structural equation are exogenous or pre-determined. You simply put all the right hand side variables in the instruments list. Can't do that! You must also know what the rank condition must be for 2SLS to be applicable and check to see if the condition is satisfied for your model. You must determine if the model should be estimated as fixed-effect or random-effect."

I sent him my eviews workfile, that's how he replies...

[startz]

the instruments MUST be the EXOGENOUS or PRE-DETERMINED variables in the structural equation plus perhaps other exogenous or pre-determined variables. You've got to decide which of the explanatory variables in the structural equation are exogenous or pre-determined.

This is the key part of your professor's question. It doesn't have anything to do with EViews. You need to have an argument from economic theory.

[bobchowz]

the instruments MUST be the EXOGENOUS or PRE-DETERMINED variables in the structural equation plus perhaps other exogenous or pre-determined variables. You've got to decide which of the explanatory variables in the structural equation are exogenous or pre-determined.

This is the key part of your professor's question. It doesn't have anything to do with EViews. You need to have an argument from economic theory.

Ok, I realize I do misunderstanding the model. Actually there is only one equation. the equation 1that FDIportion on the left hand side is actually the first stage of my TSLS...
So what now?

[startz]

the instruments MUST be the EXOGENOUS or PRE-DETERMINED variables in the structural equation plus perhaps other exogenous or pre-determined variables. You've got to decide which of the explanatory variables in the structural equation are exogenous or pre-determined.

This is the key part of your professor's question. It doesn't have anything to do with EViews. You need to have an argument from economic theory.

Ok, I realize I do misunderstanding the model. Actually there is only one equation. the equation 1that FDIportion on the left hand side is actually the first stage of my TSLS...
So what now?

My advice would be to re-read your textbook material on TSLS (or instrumental variables). It sounds like you're not understanding your professor's concerns.

[bobchowz]

My advice would be to re-read your textbook material on TSLS (or instrumental variables). It sounds like you're not understanding your professor's concerns.

appreciate your help