摘要
由于现实中的极值事件往往倾向于同时或相继发生,因此多元极值研究正成为极值统计学的理论前沿和研究热点。本文对该领域中参数建模方法的最新进展做了系统性述评,包括经典多元极值理论、Ledford-Tawn-Ramos方法和Heffernan和Tawn条件法等,并指出了这些建模方法的优缺点以及未来可能的理论突破点。本文还全面分析了近年来多元极值分析方法在金融领域的国内外应用现状,并探讨其未来的应用前景,可能是在金融传染、组合问题和系统性风险管理等方面。
Due to the frequent co-occurrences of extremes, research on multivariate extremes has become the theoretical frontier and focus in extreme value statistics. This paper systematically surveys the recent development on parametric modeling for multivariate extremes, including the classical multivariate extreme value theory, Ledford-Tawn- Ramos＇ approach and Heffernan and Tawn＇ s conditional approach, and etc. and extensively analyzes the pros and cons of these models and possible issues of theoretical improvement. Besides, we also investigate their current applications in finance, and suggest that the potential areas of future applications might be in financial contagion, portfolios and systemic risk management, and etc.
出处
《统计研究》
CSSCI
北大核心
2010年第7期65-72,共8页
Statistical Research
基金
国家自然科学基金重大国际合作研究项目(批准号:70620120444)
重点项目(批准号:70531010)
创新研究群体科学基金(70821061)
上海交通大学安泰经济与管理学院青苗基金(YK103)
关键词
多元极值
参数模型
极值统计学
极值理论
尾部相依性
Multivariate Extremes
Parametric Model
Extreme Value Statistics
Extreme Value Theory
Tail Dependence
