06-25-2017 11:19 PM
I'm trying to perform 2sls regression. The problem is that my first stage endogenous variable is binary, and the second stage dependent variable is continuous. Can I go ahead with proc logistic in first stage regressing the endogenous variable on all exogenous and IV variables, get the predicted (phat) and use OLS to regress my dependent on the exogenous and phat. Will that give unbiased and consistent estimators? I am using SAS/ETS version 14.2.