Discussion Papers, Interdisciplinary Research Project 373: Quantification and Simulation of Economic Processes 1999,94
The necessity to quantify the risk caused by the high volatility of asset prices, large insurance claims or floods has lead to an increasing interest in extreme value analysis. Generalized Pareto and extreme value distributions are well suited to model data which are exceedances above a threshold or maxima. We describe two statistical software systems - XploRe and Xtremes - that support a user in performing an extreme value analysis. Within both systems, various estimators for the above distributions are provided. We give an overview of their application and mention visual tools to check the adequacy of a parametric modeling by means of non-parametric procedures. Both systems utilize a client/server architecture to provide access to their resources across a network. While the server version of XploRe supports an interactive Java client which can be used from a web browser, the Xtremes system implements a CORBA interface that exports statistical objects to a client program.