[Limdep Nlogit List] Tobit selectivity model with endogenous regressor

[Andrea Mannberg]

Dear all,

I am trying to estimate a model with selection, inflated zeroes and
endogeneity issues. The selection equation relates to whether an individual
has had sex or not and is thus a probit. The outcome equation is a “sexual
risk index” which is only observed for individuals that are sexually active.
The inflated zeroes emanates due to the fact that a large share of the
respondents have used condoms consistently. These individuals are treated as
“zero-risk takers” in the model. The risk index is a continuous variable.
Due to the inflated zeroes, and the potential selection I’ve considered
using a Tobit model with selectivity in Limdep (I’m using Nlogit4.0).
However, one of my regressors is potentially endogenous. The problematic
regressor is a continuous measure of expected health (which is likely to be
affected by the degree of sexual risk taking). Since I am (very) new to
Limdep (and to econometrics in general) I do not know how to combine an
instrumental variable approach with the selection model. My main problem is
that I do not know how to correct the standard errors due to the generated
regressor problem.  It would be great if I could do a Murphy-Topel
correction of the standard errors, but I do not know how to code this in
Limdep. The endogenous regressor should optimally be included in the probit
and in the tobit. If any of you have any suggestions on how to do this, I
would be extremely thankful!

Best regards,

Andréa Mannberg