Estimation of Extreme Risk Regions under Multivariate Regular Variation

 

  • Profª JuanJuan Cai – Department of Econometrics and Operations Research – University of Tilburg
  • FCUL (DEIO) – Campo Grande – Bloco C6 – Piso 4 – Sala 6.4.30 – 15h
  • Segunda-feira, 24 de Maio de 2010
 
 Abstract: When simultaneously monitoring d possibly dependent risks one is often interested in extreme risk regions, with very small probability p. We consider regions of the form {f<c}, where f is the joint density. We assumethat our data are i.i.d. d-dimensional regularly varying random vectors. Using extreme value theory, we construct a natural estimator of such aregion and prove a refined form of consistency. We also present a simulation study and an application on foreign exchange rate data.