Dear statalists,
I am very new to STATA and trying to use 2SLS when the IV is a binary variable. I have gone through the literature and it supports the use of 2SLS with first-stage probit model. Specifically, what I am required to do are:
1. Run a probit model for panel data: I used command
Code:
xtprobit AI X1 X2 X3 X4
. But, I would like to concern the Year Fixed effects, and also Type-specific fixed effects of individuals (In my case, that is industry) and NO entity fixed effects. What is the option that I should use? pe or re? And, how can I include fixed effects for xtprobit?
2. Generate the fitted probability of AI: I used command:
Code:
predict PAI
. It shows the xb is defaulted. Am I am not quite certain that xb (linear prediction) is what I need in this phase. Any advice to overcome this?
3. Use the fitted probability of AI, PAI, to run a 2SLS for panel data: I used the command:
Code:
xtivreg2 DepVar AI IndeVar
. But I am would like to use ONLY time fixed effects, industry fixed effects, and NO entity fixed effects here! What is the exact code that I should use?

Thanks so much for your reading. Hope to learn from you!
Best regards,
Roman