Large deviations of bootstrapped \(U\)-statistics (Q943614)

This paper develops a large deviation result with Cramér series for bootstrapped U-statistics where the U-statistic has a non-degenerate, bounded kernel. The method of proof is based on the contraction technique of \textit{R. Keener, J. Robinson} and \textit{N.C. Weber} [Stat. Probab. Lett. 37, No. 1, 59--65 (1998; Zbl 0928.60016)]. The result gives the best possible remainder term and, in particular, provides information on the accuracy of the bootstrap approximation as a relative error in the large deviation case analogous to the result for the bootstrapped sample mean given by \textit{P. Hall} [J. Multivariate Anal. 35, No. 1, 108--129 (1990; Zbl 0705.62045)].