On Resampling Inference in Econometric Models

Abstract

This paper examines alternative applications of resampling methods to estimation and inference in econometric models. The major applications of resampling have been in the construction of sampling distributions. Here, we demonstrate that in models specified by moment condition, often referred to as generalised method of moments, resampling can be used to generate testable implications of these moment restrictions. We also use resampling techniques to correct for the bias of instrumental variables estimation. This estimates the leading term in the Nagar expansion for the bias of such an estimator. Monte Carlo experiments suggest that the correction can lead to sharp improvements in the bias of such estimators in small samples.KeywordsInstrumental VariableLinear Regression ModelMoment ConditionBias CorrectionVariance Covariance MatrixThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.