MENU
| Título | Nonstationary extremes and the US business cycle |
| Publication Type | Unpublished |
| Year of Publication | 2010 |
| Authors | de Carvalho M, Turkman FK, Rua A |
| Series Title | Preprint |
| Palavras-chave | Box-Jenkins method, Generalized Pareto distribution, Nonstationary process, Peaks over threshold, statistics of extremes, Unemployment data, US business cycle |
| Abstract | Considerable attention has been devoted to the statistical analysis of extreme events. Classical peaks over threshold methods are a popular modelling strategy for extreme value statistics of stationary data. For nonstationary series a variant of the peaks over threshold analysis is routinely applied using covariates as a means to overcome the lack of stationarity in the series of interest. In this paper we concern ourselves with extremes of possibly nonstationary processes. Given that our approach is, in some way, linked to the celebrated Box-Jenkins method, we refer to the procedure proposed and applied herein as Box-Jenkins-Pareto. Our procedure is particularly appropriate for settings where the parameter covariate model is non-trivial or when well qualified covariates are simply unavailable. We apply the Box-Jenkins-Pareto approach to the weekly number of unemployment insurance claims in the US and exploit the connection between threshold exceedances and the US business cycle. |
| URL | http://www.dm.fct.unl.pt/sites/www.dm.fct.unl.pt/files/preprints/2010/5_10.pdf |