This paper presents a unified second order asymptotic framework for conducting inference on parameters of the form $\phi(\theta_0)$, where $\theta_0$ is unknown but can be estimated by $\hat\theta_n$, and $\phi$ is a known map that admits null first order derivative at $\theta_0$. For a large number of examples in the literature, the second order Delta method reveals a nondegenerate weak limit for the plug-in estimator $\phi(\hat\theta_n)$. We show, however, that the "standard" bootstrap is consistent if and only if the second order derivative $\phi_{\theta_0}''=0$ under regularity conditions, i.e., the standard bootstrap is inconsistent if $\phi_{\theta_0}''\neq 0$, and provides degenerate limits unhelpful for inference otherwise. We thus identify a source of bootstrap failures distinct from that in Fang and Santos (2015) because the problem persists even if $\phi$ is differentiable. We show that the correction procedure in Babu (1984) can be extended to our general setup. Alternatively, a modified bootstrap is proposed to accommodate nondifferentiable maps. Both approaches are shown to provide local size control under restrictions on $\hat\theta_n$ and $\phi_{\theta_0}''$. As an illustration, we develop a test of common conditional heteroskedastic (CH) features that allows the existence of multiple common CH features. In fact, our paper contains new results on the $J$-test in GMM settings that allow partial identification and/or degeneracy of Jacobian matrices.
Read full abstract