Abstract
The main limitation of many statistical methods lies in the fact that they usually require fulfilments of many different assumptions that cannot often be verified. This problem arises frequently in analysis of time series, too. In this case it is helpful to use bootstrap methods. This paper deals with the application of the bootstrap method to the Box-Jenkins methodology for reduction of bias. Several methods of nonparametric boot-strap for a bias-reduced estimate of the autoregressive parameters of AR(1) and AR(2) are presented in this paper : covering model oriented bootstrap, overlapping moving blocks and not-overlapping moving blocks. A comparison of the results of the classical and resampling methods is performed for the premium written time series.
Keywords
Get full access to this article
View all access options for this article.