报告地点：行健楼学术活动室526
邀请人：孙红芳 博士
Abstract: We propose a Double Pooling Dynamic Tail Estimation (DPDTE) framework for jointly estimating time-varying tail risks in panel data. The method directly recovers multiple Value-at-Risk (VaR) levels and the associated extreme Expected Shortfall (ES) through a unified loss function that pools information across both cross-sectional units and quantile levels. Tail dynamics are modeled using a score-driven specification in combination with extreme value theory. We establish consistency and asymptotic normality of the estimators and document favorable finite-sample properties in simulations. Applying DPDTE to Growth-at-Risk (GaR) and extreme ES for 14 European economies from 1985 to 2025, we uncover highly persistent euro-wide tail factors, substantial cross-country heterogeneity in long-run downside exposure, and marked differences between tail-driven and location-driven downturns during the global financial crisis, the euro-sovereign episode, and the COVID-19 shock. These findings highlight the value of double pooling for identifying extreme macroeconomic risks in the presence of sparse tail information and ragged-edge data structures.
报告人简介： 叶仕奇，男，经济学博士。中国科学院数学与系统科学研究院预测科学研究中心，助理研究员。从事高维复杂系统建模、宏观经济风险管理、大数据方法与应用等领域的研究工作。相关成果发表于《经济研究》《中国工业经济》《世界经济》《统计研究》《数量经济技术经济研究》、Journal of Economic Dynamics and Control、Journal of Management Science and Engineering、Energy Economics等国内外学术期刊。 主持首批国家自然科学基金青年学生基础研究（博士研究生）项目以及博士后面上资助、特别资助、国资计划等项目。获得首届中国宏观经济学会优秀论文奖、福建省社会科学优秀成果奖、澳大利亚新西兰计量经济学研究组（ANZESG）青年计量经济学家奖（最佳论文报告奖）等国内外奖项。
