Hi! I have been trying to estimate the relationship between an endogenous variable (say X) and Y, and X and Z using the instrumental variable approach. I have panel data. So I have used 'xtivreg' command with 'fe' for separate estimations X and Y, and X and Z; in these regressions, I have instrumented the endogenous variable, X, with an IV, say B. Now, I think variables Y and Z are also predictors of each other. So I think I can use 3SLS in this case, can't I? If so, could someone suggest to me how to incorporate 'fe' in 3SLS in STATA?
-
Login or Register
- Log in with
