Hi everyone,

I am trying to estimate a double hurdle model: Probit and lognormal (Craggs model), I would like to address the issue of endogeneity with one of my independent variables. I have seen some studies use an IV and others use a control function approach. The issue is my variable of interest appears only in the lognormal portion of the double hurdle. Would conventional IV approaches suffice by regressing my independent variable of interest in a first-stage regression with the IV and other controls and using the predicted values in my double hurdle model?