st: Dealing with multiple endogenous variables in Stata 12 for mlogit

[Rieza Soelaeman <rsoelaeman@gmail.com>]

Dear Statalisters,
I am trying to fit a model with several endogenous variables where the
outcome is a categorical variable with >2 alternatives.  For illustrative
purposes, the model I'm trying to fit is:

care_ij = X_ij + num_fac_j + insurance_ij

Where "care_ij" is the type of care sought (no care, public facility,
private facility); X_ij is a vector of individual-level exogenous
variables; num_fac_j is a potentially endogenous continuous variable on the
number of health care facilities in a geographic area (say regency) where
the individual lives; and insurance_ij is a potentially endogenous variable
on whether the individual is covered by insurance X.

The papers I've read (e.g. Angeles, Guilkey, and Mroz, 1998) simultaneously
modeled the endogenous variables along with the outcome.  I believe some of
these were done in Fortran.  Is there a way to do this in Stata 12, aside
from manually predicting* the values of num_fac_j and insurance_ij using
all of the instruments and using the predicted values in the structural
model?  In my actual model, I have 3 endogenous continuous variables,
and there are 4 insurance schemes, rather than a simple insurance vs.
no insurance.


Thanks,
RHS


*so for example getting the predicted values of these:
num_fac_j = expenditure_j + region_j + Xij+ civil_servant_ij
insurance_ij=civil_servant_ij + Xij + expenditure_j + region_j
(where "expenditure" and "region" are instruments for # of facilities and
"civil servant" is an instrument for insurance)

and then using mlogit as foll:
mlogit care_ij X_ij num_fac_j_hat insurance_ij_hat, vce(robust)
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/