I encountered a project which has binary outcomes and repeated observations of the same individual,
but the panel is short (
T is small), so it seems that the only choice is conditional logit (
clogit in Stata). One of the regressors, say,
x1 is endogenous, and I construct a solid IV (instrumental variable)
z1 to instrument it.
The question is: is there an estimation method that can run fixed-effect non-linear iv estimation?
Thanks so much for your time and knowledge