Hello,
I have gone through multiple posts/replies on Statalist regarding endogeneity but I am still stuck so pardon my ignorance here. I would very much appreciate help on this-
I have a panel dataset (firms and years). I have 5 independent variables X1-X5. X1-X3 are the controls and I have an interaction term as well. All variables are continuous. X1-X3 and Y vary with time. X4 and X5 do not vary with time.
I am running an OLS regression with double-clustered robust standard errors using the -IVREG2- command in Stata as shown below-
ivreg2 Y X1 X2 X3 X4 X5 X4*X5 , cluster(firm years)
I want to
1) check if X3 and X4 are possibly endogenous.
2) If they are, then I want to use -IVREG2H- to generate instruments for them.
As you will notice that I also have an interaction term involving the 2 potential endogenous variables. How do I test and correct for endogeneity in such a case?
ivreg2h Y X1 X2 X3 (X4 X5 X4*X5=),cluster(firm years) gmm2s <- something like this? However, in this case, I am basically considering the interaction of X4 and X5 as an altogether new endogenous variable which is not making sense to me.
Thank you so much in advance!
I have gone through multiple posts/replies on Statalist regarding endogeneity but I am still stuck so pardon my ignorance here. I would very much appreciate help on this-
I have a panel dataset (firms and years). I have 5 independent variables X1-X5. X1-X3 are the controls and I have an interaction term as well. All variables are continuous. X1-X3 and Y vary with time. X4 and X5 do not vary with time.
I am running an OLS regression with double-clustered robust standard errors using the -IVREG2- command in Stata as shown below-
ivreg2 Y X1 X2 X3 X4 X5 X4*X5 , cluster(firm years)
I want to
1) check if X3 and X4 are possibly endogenous.
2) If they are, then I want to use -IVREG2H- to generate instruments for them.
As you will notice that I also have an interaction term involving the 2 potential endogenous variables. How do I test and correct for endogeneity in such a case?
ivreg2h Y X1 X2 X3 (X4 X5 X4*X5=),cluster(firm years) gmm2s <- something like this? However, in this case, I am basically considering the interaction of X4 and X5 as an altogether new endogenous variable which is not making sense to me.
Thank you so much in advance!