【专家简介】：洪永淼，中国科学院数学与系统科学研究院关肇直首席研究员，中国科学院大学经济与管理学院院长，发展中国家科学院院士，世界计量经济学会会士，亚太人工智能学会会士，亚洲金融经济研究局高级会士，教育部高等学校经济学类专业教学指导委员会副主任委员。曾任美国康奈尔大学经济学与国际研究讲席教授、统计学教授，中国留美经济学会会长。
研究领域为计量经济学、时间序列分析、金融计量学、统计学，在Annals of Statistics、Biometrika、Econometrica、Journal of American Statistical Association、Journal of Political Economy、Journal of Royal Statistical Society B、Management Science、Quarterly Journal of Economics、Review of Economic Studies、Review of Financial Studies、《经济研究》《管理世界》《中国工业经济》《管理科学学报》《中国科学院院刊》等经济学、金融学和统计学中英文主流期刊发表文章180余篇。出版《Python经济大数据分析》《概率论与统计学》《高级计量经济学》、Probability and Statistics for Economists、Foundations of Modern Econometrics: A Unified Approach等中英文著作。2014-2025年连续12年入选Elsevier经济学/统计学中国高被引学者榜单，获2022年高等教育（本科）国家级教学成果奖一等奖。
【报告摘要】：Detecting extreme risk spillovers is important for understanding how rare but consequential shocks propagate across time series. The cross-spectrum provides a natural tool for investigating Granger causality between time series and is therefore well suited to spillover testing. Existing cross-spectral tests for Granger causality in risk typically treat the VaR risk levels as fixed constants. However, when the risk levels of interest are close to 0 (or 1), it is more appropriate to allow them to move gradually toward the tail as the sample size increases. In this case, the extreme quantile parameter estimators converge at a rate slower than T, thereby introducing greater parameter estimation uncertainty into the original test statistic and distorting its asymptotic null distribution. In this paper, we propose a class of improved kernel-based cross-spectral tests. Under the improved testing framework, the impact of greater parameter estimation uncertainty is removed by adopting the Wooldridge (1990) device. Consequently, our framework allows the risk levels to approach 0 (or 1) as the sample size tends to infinity, and thereby accommodates the use of estimators from extreme quantile regression models. We prove that under appropriate regularity conditions, the improved tests remain asymptotically standard normal under the null hypothesis. Simulation results show that the improved tests exhibit more reasonable finite-sample size than the original tests under fixed risk levels, and maintain good performance along several shrinking paths for the risk levels as the sample size increases. The empirical findings from testing for extreme risk spillovers between the intraday gold market and a set of foreign exchange markets highlight the practical value of our approach.
【报告时间】：2026年06月26日（周五）9:00-10:00
【报告地点】：位育楼417
