The Stata Journal has published an article by Christopher F. Baum, Mark E. Schaffer, and Steven Stillman, which extends their 2003 paper on instrumental variables (IV) and generalized method of moments (GMM) estimation. The paper introduces enhanced routines for IV/GMM estimation and testing, including robust standard errors, tests for endogeneity, weak instruments, and autocorrelation. The authors describe the analytical underpinnings of these enhancements and provide a detailed overview of the new features in the ivreg2 command. The paper also discusses the use of GMM in time-series contexts and the implementation of heteroskedasticity and autocorrelation-consistent (HAC) standard errors. The authors also present examples of how to use the new features in practice, including the use of HAC standard errors in a quarterly time-series model. The paper concludes with a discussion of the performance of different estimation methods, including CUE, LIML, and k-class estimators, and their implications for finite-sample inference. The authors emphasize the importance of testing for underidentification and weak identification in IV/GMM estimation and provide guidance on how to interpret the results of these tests. The paper also discusses the use of the Frisch-Waugh-Lovell theorem and the implications of a rank-deficient covariance matrix of orthogonality conditions. Overall, the paper provides a comprehensive overview of the latest developments in IV/GMM estimation and testing, with practical examples and guidance on their implementation in Stata.The Stata Journal has published an article by Christopher F. Baum, Mark E. Schaffer, and Steven Stillman, which extends their 2003 paper on instrumental variables (IV) and generalized method of moments (GMM) estimation. The paper introduces enhanced routines for IV/GMM estimation and testing, including robust standard errors, tests for endogeneity, weak instruments, and autocorrelation. The authors describe the analytical underpinnings of these enhancements and provide a detailed overview of the new features in the ivreg2 command. The paper also discusses the use of GMM in time-series contexts and the implementation of heteroskedasticity and autocorrelation-consistent (HAC) standard errors. The authors also present examples of how to use the new features in practice, including the use of HAC standard errors in a quarterly time-series model. The paper concludes with a discussion of the performance of different estimation methods, including CUE, LIML, and k-class estimators, and their implications for finite-sample inference. The authors emphasize the importance of testing for underidentification and weak identification in IV/GMM estimation and provide guidance on how to interpret the results of these tests. The paper also discusses the use of the Frisch-Waugh-Lovell theorem and the implications of a rank-deficient covariance matrix of orthogonality conditions. Overall, the paper provides a comprehensive overview of the latest developments in IV/GMM estimation and testing, with practical examples and guidance on their implementation in Stata.