A `Nonparametric Plug-In Rule' for Selecting Optimal Block Lengths
for Block Bootstrap Methods
Soumendra N. Lahiri, (Iowa State University), firstname.lastname@example.org
In this talk, we consider the problem of choosing the optimal block size for block bootstrap methods empirically. We suggest a general plug-in principle for estimating MSE-optimal smoothing parameters, and establish its validity for estimating the of the optimal block size for block bootstrap methods. Unlike standard plug-in rules, the proposed method can be applied without explicit analytical expressions for the constants that appear in the leading term of the optimal block size. We also discuss computational efficacy of the method and illustrate its finite sample properties through numerical examples.