Bootstrap refinements for GMM based tests (in Russian)

This essay provides a brief review about bootstrap higher order refinements for tests based on generalized method of moments estimators. First, we briefly describe the asymptotic behavior of two-step GMM estimators. Second, we give a heuristic argument for why inference based on bootstrap critical values is more accurate than that based on asymptotic normality. Third, we briefly summarize nonparametric resampling methods. Fourth, we outline how critical values based on the block bootstrap reduce the error in the rejection probability for t-tests based on GMM estimators. Finally, we give a overview of some alternative bootstrap procedures which provide improvements over the block bootstrap refinements.