报告时间：2026年11月19日，周四下午2:00-3:00
地点：36-316
报告人: 李睿 上海对外经贸大学
摘要: In this paper, we propose a single-index varying-coefficient panel data model that incorporates the unobservable multiple interactive fixed effects, where such fixed effects are permitted to be correlated with the covariates. We fuse B-spline approximation into the least squares estimation paradigm, and implement a robust iterative algorithm to acquire the estimates of the link function and all coefficient functions. Under a set of mild regularity conditions, we establish the consistency and asymptotic normality of the proposed estimators. Furthermore, we construct two generalized likelihood ratio statistics based on the residual-based block bootstrap procedure, to respectively test whether the link function is linear or nonlinear, and whether the coefficient functions are constant or non-constant. To mitigate the influence of block length selection on the performance of hypothesis testing, a Warp-speed approach is introduced to determine the optimal block length. Extensive Monte Carlo simulations are conducted to display the favorable finite-sample performance of our proposed estimation and inference procedures. Finally, we apply the proposed approach in some real data analyses.
报告人简介：上海对外经贸大学统计与数据科学学院副院长，教授，博士生导师。上海财经大学统计学博士。2016年入职上海对外经贸大学。2019-2020年到新加坡国立大学访问交流。主要从事统计与数据科学的理论、方法和应用研究，主持国家社会科学基金一般项目、教育部人文社会科学研究规划基金项目、全国统计科学研究重点项目、教育部重点实验室项目和上海市科学技术委员会项目等，在统计学SCI权威期刊JCGS、JBES、Stat. Sin.、SJS、JMA、CSDA、INS、NN等发表学术论文近40篇。担任中国现场统计研究会大数据统计分会常务理事，全国工业统计学教学研究会民族统计与数据科学分会常务理事等。
