报告题目：ExtremeContinuousTreatmentEffects: Measures,EstimationandInference
主讲人：彭柳华（墨尔本大学）
报告时间：2023年12月18日16:00-17:00
报告地点：beat365平台会议室（文波401）
摘要:This paper concerns estimation and inference for treatment effects on deep tails of the potential outcome distributions. For a tail level
, we consider two measures for tail characteristics: the th-quantile and the th-tail mean defined as the conditional mean beyond the th-quantile. Then for that is close to 0 or 1, we define the extreme quantile treatment effect (EQTE) and the extreme average treatment effect (EATE), which are the differences of their corresponding measures at different treatment status. The EQTE and EATE are identified through the commonly adopted unconfoundedness condition and are estimated with the aid of extreme value theory. Our limiting theory is for the EQTE and EATE processes indexed by a set of tail levels and hence facilitates uniform inference. We propose a novel bootstrap procedure for finite-sample implementation. Our theory is focused on the continuous treatment effect model, but can be readily extended to the discrete model which is technically easier to handle. Simulations suggest that our method works well in finite samples and an empirical application illustrates its practical merit.
主讲人简介：彭柳华博士，2017年获得爱荷华州立大学（IowaStateUniversity）统计系统计博士学位。现受聘为墨尔本大学（theUniversityofMelbourne）数学与统计学院高级讲师（等同于副教授）。彭博士的主要研究领域为高维数据分析，极值理论，大数据的分布式推断，bootstrap，以及非参数假设检验。彭博士在theAnnalsofStatistics，JournalofRoyalStatisticalSocietySeriesB（JRSSB），Biometrika，EconometricTheory，StatisticaSinica等国际顶尖统计与计量期刊发表论文十余篇。彭博士多次应邀为顶尖统计期刊专家评审及在国际会议做邀请报告。
