2SLS estimation

[John Economou]

Hello,
I would like to estimate the following 2SLS panel regression:

First stage: Z_it = α W_it*F_it + γ F_it + k_i + m_t + u_it

Second stage: Y_it = β Z_(hat)it + δ F_it + k_i + m_t + e_it

where Z_(hat) is the predicted Zhat.

---

I am using Stata 15.1.
My first thought is to use xtivreg2, like that:

gen wf = w*f // create the interaction variable
xi: xtivreg2 y (z = wf f) f i.year, first fe ro

however, my concern is that by using the above command I am not take into account the predicted value of Z (Zhat) of the second stage.

Could you please give some advices...
Thanks a lot,
John.

[Joro Kolev]

Both -xtivreg- and -xtivreg2- do what you want to do.