Seminar Notice
 
Statistical Laboratory
Iowa State University
 
DATE AND TIME: Friday, January 21, 2005, 4:10 p.m.
 
PLACE: 319 Snedecor
 
SPEAKER: Liang Peng, School of Mathematics, Georgia Institute of Technology, Atlanta, Georgia
                                            
TITLE: Data Tilting for Rare Events
 

ABSTRACT

 
Rare events like floods, earthquakes, wild fires, etc., always incur tremendous losses to our society.  For estimating the low probabilities of those rare events, it involves extrapolating data in general.  Based on the assumption on the normalized maxima, extreme value theory offers the ability to extrapolate data, and has been applied to estimate extreme tail probabilities and high quantiles.  In this talk, we will present methods for constructing confidence intervals for high quantiles based on either iid data or GARCH models.  One appealing method is the data tilting method proposed by Hall and Yao (2003), which is considered as a generalization of empirical likelihood method.  We study those methods both theoretically and empirically.
 
Real applications are involved too.  In particular, we argue that interval estimation for high quantiles is more important than point estimation in the context of risk management.
 
COFFEE: 3:45 p.m., 104 Snedecor Hall