主 题：Tail dependence, extreme quantile and predictive regression
主讲人：Liang Peng教授
主持人：龚金国
时 间：2014年7月2日上午9：00
地 点：通博楼B212学术会议室
主办单位：统计学院 科研处
主讲人简介：
Liang Peng，复旦大学管理学院教授，Georgia Institute of Technology教授。主要研究方向为：Extreme value theory and risk analysis in insurance and finance, Heavy tailed, long-range dependent, nonlinear time series, Nonparametric smoothing and empirical likelihood methods, Financial Econometrics, Copulas and actuarial sciences.已在STAT SCI,ANN STAT，JASA, Econometric Theory等著名杂志发表多篇论文。
内容提要：
First we will introduce extreme value theory and a new approach for distinguishing asymptotic dependence and asymptotic independence which plays an important role in predicting rare events. Second we propose a new semi-parametric approach to estimate high quantiles, where both parametric and nonparametric methods are not appropriate. Finally we propose a way of uniform interval estimation for predictive regressions regardless of the predicting variable being stationary or near integrated or having an infinite variance.
