# Heckman/Two part model with endogenous binary variables

I have a model with the continous variable(Y) that takes zero values. So, I am trying to use the limited dependent variable model (Two part/Heckman) , but I have binary endogeneous variables (two) ($X_1$ and $X_2$) also which I think are jointly determined. I am guessing now I have three probit (including probit part of the heckman(Two part). So, this is the trivaariate probit model. Am I in the right path? or is there any standard modelling procedure to handle this issue.

## 1 Answer

Try user-written cmp in Stata.

• Thanks. I have already looked into cmp. My question at this point is, whether I can extend biprobit +ols for y (with selection adjustment coefficients) version as discussed in Madalla (p.278) or equation 19 of this paper as triprobit plus ols for y (with selection adjustment coefficients). My model have two endogenous vars and one probit from two part(heckman), hence the need for trivariate probit. Also I need to model continuous var y with the selection adjustment coefficients as in biprobit case. Any reference to the papers would be appreciated. – user227710 Feb 17 '14 at 23:41