Document Search
Add To My Bookshelf Sign in or Register Save And Annotate
Keywords:

EVT sign in to follow this
extreme sign in to follow this
Pareto sign in to follow this
multivariate sign in to follow this
shortfall sign in to follow this
Categories:

VaR Methods sign in to follow this
--Extreme Value Theory sign in to follow this
Half-Life:
Impact:
Discuss This Paper
Sign in to follow this page
Recent Comments
  more
On Multivariate Generalised Pareto Distributions and High Risk Scenarios
Company: ETH Zurich
Year Of Publication: 2006
Month Of Publication: March
Pages: 111
Download Count: 477
View Count:
Comment Num: 0
Language: EN
Source:
Who Can Read: Free
Date: 12-28-2006
Publisher: Administrator
Summary
AbstractConsider a large investor’s market position, which may comprise a several hundredassets. Such a market position can be thought of as a d-dimensional vector, whosecomponents are the prices or log-prices of the individual assets. Risk managers areinterested in the value of the position at some fixed future date, say one or ten tradingdays ahead—a customary time horizon in market risk management in banking. Whilethe present position is known, z0 2 Rd, the future position is unknown and may bedescribed by a d-dimensional random vector Z whose distribution is concentratedaround the value z0.A major concern in financial risk management is that the market drifts off inan undesirable direction. Put differently, situations where Z will lie in some riskyregion far away from z0 raise worries. In order to apply standard risk managementtools such as the calculation of Value-at-Risk (VaR) or expected shortfall (ES) it isdesirable to have a multivariate model for the random vector Z.Amongst others, parts of a theory developed by Balkema and Embrechts [6] arediscussed in this thesis. The approach was developed to suit a setup as describedabove. In univariate extreme value theory (EVT), the concept of exceedances overthresholds has been soundly studied and is well understood. According to the wellknownPickands-Balkema-de Haan Theorem, the distribution of appropriately scaledexceedances over high thresholds is well approximated by a generalised Pareto distribution(GPD). The extension of this concept to the multivariate case is howevernot straightforward, mainly due to the lack of a natural ordering relation. Althoughthere has been much less research, some useful approaches emerged. Tajvidi’s [36] aswell as Balkema and Embrechts’ [6] contributions are discussed more extensiv
Author(s)
Degen, Matthias Sign in to follow this author
This document's citation network:
Similar Documents:
Close window
Sign up in one step, no personal information required. Already a Member?



Email:
Repeat Email:
User Name:
Password:
Confirm Password:

Sign Up


Welcome to GloriaMundi!
Thanks for singning up



continue or edit your profile