I discuss the implementation of two state-of-the-art econometric estimators in Stata. The first addresses the problem of estimating a binary response with one or more limited endogenous variables. Lewbel, Dong, and Yang (Canadian J. Econ., 2012) present a solution based on Lewbel's "special regressor" method, a version of which I have implemented in Stata as sspecialreg. The second considers the problem of estimating an equation with instrumental variables techniques where sufficient instruments may not be available. Lewbel (J. Bus. Econ. Stat., 2012) presents a solution based on generated instruments that take advantage of heteroskedasticity. I have implemented that routine as an extension of Baum, Schaffer, and Stillman's ivreg2 (Stata J., 2003, 2007) as ivreg2h. I will motivate these two extensions to Stata's econometric capabilities with illustrations of their use.
