Some benefits of standardisation for conditional extremes

TL;DR

This paper discusses the benefits of standardising marginals in multivariate extreme value models, particularly within the Heffernan and Tawn (2004) framework, to address theoretical issues highlighted by counterexamples.

Contribution

It demonstrates that standardising marginals in the Heffernan and Tawn model removes problems identified by counterexamples, strengthening the theoretical foundation.

Findings

Standardisation removes counterexample issues.

Supports the validity of the Heffernan and Tawn approach.

Enhances confidence in multivariate extreme value modeling.

Abstract

A key aspect where extreme values methods differ from standard statistical models is through having asymptotic theory to provide a theoretical justification for the nature of the models used for extrapolation. In multivariate extremes many different asymptotic theories have been proposed, partly as a consequence of the lack of ordering property with vector random variables. One class of multivariate models, based on conditional limit theory as one variable becomes extreme, developed by Heffernan and Tawn (2004), has developed wide practical usage. The underpinning value of this approach has been supported by further theoretical characterisations of the limiting relationships by Heffernan and Resnick (2007) and Resnick and Zeber (2014). However Drees and Jan{\ss}en (2017) provided a number of counterexamples of their results, which potentially undermine the trust in these statistical methods. Here we show that in the Heffernan and Tawn (2004) framework, which involves marginal standardisation to a common exponentially decaying tailed marginal distribution, the problems in these examples are removed.