Standard-error Correction in Two-stage Optimization Models: A Quasi–maximum Likelihood Estimation Approach

摘要

Following Wooldridge (2014), we discuss and implement in Stata an efficient maximum likelihood approach to the estimation of corrected standard errors of two-stage optimization models. Specifically, we compare the robustness and efficiency of this estimate using different non-linear routines already implemented in Stata such as ivprobit, ivtobit, ivpoisson, heckman, and ivregress.