Oracle Inference for GMM Models

• Activity Number: 296
• Type: Topic Contributed
• Date/Time: Tuesday, August 6, 2013 : 8:30 AM to 10:20 AM
• Sponsor: ENAR
• Author(s): Mihai Giurcanu*+ and Brett Presnell
• Companies: University of Florida and University of Florida
• Keywords: GMM inference ; misspecification ; oracle estimation ; standard bootstrap ; oracle bootstrap ; spatial models

Abstract

The Generalized Method of Moments (GMM) is a popular estimation technique in Econometrics which is used to fit models in which the number of parameters is exceeded by the number of moment conditions identifying them. In our effort to allow that some of the moment conditions may be misspecified, we develop an oracle GMM estimator which automatically selects the true moment conditions for large sample sizes. We show that the standard bootstrap estimators of the null distributions of the Wald, likelihood-ratio type, score-type, and J tests are inconsistent in this setting. However, extending the oracle bootstrap developed by Giurcanu and Presnell (2009), we develop consistent bootstrap tests for the oracle GMM estimators. The results of a simulation study and the data analysis of a spatial dynamic panel data model show the finite sample properties of the bootstrap tests and the application of our oracle bootstrap methodology in the GMM context, respectively.