题目：A jackknife score test with very many weak instruments
报告人：郭旭
报告时间：2026年3月20日 15:00
地点：综合楼644会议室
报告人简介：
郭旭，现任北京师范大学统计学院教授，博士生导师。主持国家自然科学基金青年基金B类。曾荣获北京师范大学第十一届“最受本科生欢迎的十佳教师”，北京师范大学第十八届“青教赛”一等奖和北京市第十三届“青教赛”三等奖。目前主要关注高维回归模型中的假设检验问题也对基于机器学习算法的统计推断感兴趣，有多篇文章发表在统计学和计量经济学国际顶尖期刊，担任统计学国际知名期刊JMVA副主编。
报告摘要：
The inference of instrumental variable regression models with many weak instruments has drawn much attention. However, many existing methods restrict the number of instrumental variables to be less than the sample size. In this paper, we propose a new jackknife score test which is still valid with many instruments, where the number of instruments is allowed to be larger than sample size, weak instruments and heteroskedastic errors. The asymptotic properties of our jackknife score test are derived under the null and alternative hypotheses. Moreover, we establish the joint asymptotic normality of our jackknife score test and a newly introduced jackknife Anderson-Rubin test. We further consider a Cauchy combination test to integrate our jackknife score test and the newly introduced jackknife Anderson-Rubin test. Numerical and real data analytical results show that the two tests proposed in this paper effectively control the empirical size and achieve desirable test power across diverse settings.
