Correct specification for IV when variables are interacted
Posted: Wed Sep 20, 2017 5:45 pm
Hello All
I very much appreciate your help on the following. My regression is Y=a0+a1*x1+a2*x1*x2+controls (simplified). I believe that x1 is endogenous and would like to estimate the regression using the 2SLS with the instrumental variable approach. I found Z that should be a good instrument. Now, What should I put in the Instruments list when I estimate 2SLS? I put "Y then c then Z then Z*x2 then controls". Is this a correct specification? I am asking because in my original regression, x1 (endogenous) is interacted with x2 that is exogenous. So, would it be correct to specify "Z then Z*x2" in the instruments list? Thank you!!!!!
I very much appreciate your help on the following. My regression is Y=a0+a1*x1+a2*x1*x2+controls (simplified). I believe that x1 is endogenous and would like to estimate the regression using the 2SLS with the instrumental variable approach. I found Z that should be a good instrument. Now, What should I put in the Instruments list when I estimate 2SLS? I put "Y then c then Z then Z*x2 then controls". Is this a correct specification? I am asking because in my original regression, x1 (endogenous) is interacted with x2 that is exogenous. So, would it be correct to specify "Z then Z*x2" in the instruments list? Thank you!!!!!