“Simpler standard errors for two-stage optimization estimators” (Terza, 2016a, Stata Journal 16: 368–385) offers an analytic simplification of the daunting textbook formulations of the asymptotic variance–covariance matrix of a class of two-stage optimization estimators. Here I revisit that simplification and show that it applies to a much broader class of estimators than was originally considered. I also offer a correction that further enhances the generality of this asymptotic variance–covariance matrix formulation. These points are illustrated via a real-data application.
MullahyJ.1997. Instrumental-variable estimation of count data models: Applications to models of cigarette smoking behavior. Review of Economics and Statistics79: 586–593. https://doi.org/10.1162/003465397557169.
TerzaJ. V.2016b. Inference using sample means of parametric nonlinear data transformations. Health Services Research51: 1109–1113. https://doi.org/10.1111/1475-6773.12494.
TerzaJ. V.2023a. Simpler standard errors for two-stage M-estimators in empirical economics using numerical derivatives. Unpublished manuscript.
6.
TerzaJ. V.2023b. Standard errors for regression-based causal effect estimates in economics using numerical derivatives. Unpublished manuscript.
7.
TerzaJ. V.BasuA.RathouzP. J.2008. Two-stage residual inclusion estimation: Addressing endogeneity in health econometric modeling. Journal of Health Economics27: 531–543. https://doi.org/10.1016/j.jhealeco.2007.09.009.
