|
Abstract:
|
The block bootstrap is known to provide valid inference for weakly dependent time processes and, in this context, much research has focused on determining the large-sample properties of block bootstrap approximations for sample means. This talk discusses a block bootstrap method for a practically broad class of linear processes exhibiting strong dependence, which includes popular models for long-memory. While a great deal of research has focused on block bootstrap variance estimators for the sample mean in the weak dependence case, little has been known about block bootstrap performance under strong dependence. We provide a detailed examination of the bias and variance of block bootstrap estimators under long-range dependence, which differ critically from the weak dependence setting. The findings are illustrated with simulation.
|
