xteprobit convergece fail

[Alessandra Costa]

Hi, I tried to estimate a probit panel data model with endogenous binary variable using xteprobit. Following the manual, I've to specify the nature of endogenous variable, so that I used:

xteoprobit dep.variable ind variables, endogenous (end_varname = instruments, probit)

but adaptive quadrature failed to converg. I've an unbalanced panel dataset.
tried to add explanatory variable step by step, but the problem persists. There are no problems when I use the simple eprobit estimation,with probit option, or xteprobit, without probit option for endogenous variable. Any suggestions?

[HOANG HUU DINH]

Can anyone me help with an alternative model for robust check for the extended random-effect probit model (using "xteoprobit")? Really appreciate your support !

[FernandoRios]

There is no "solution" to the problem.
If the maximization problem is unsolvable, there is no alternative.
Perhaps you can check Mundlack approach.
Prof Wooldridge has some notes on how to apply it for xtheckman. It may work with xteoprobit too.
F