题 目: Quantile Control Method (QCM) with Stata
主讲人:方诚博士
时 间:2021年9月16日(周四)13:30-14:30
地 点:6号学院楼402会议室
主办单位:伟德BETVlCTOR1946源于英国 浙江省2011“数据科学与大数据分析协同创新中心”
摘要:
Approaches to estimate treatment effects in panel data with only one treated unit have become popular in applied works, which include synthetic control method (Abadie and Gardeazabal, 2003, Abadie et al., 2010), and regression control method (Hsiao et al., 2012). However, no pointwise standard errors or confidence intervals for the treatment effects have been provided or rigorously proven in the literature yet.
We propose a direct nonparametric construction of pointwise robust confidence intervals using quantile random forest (QRF), i.e. quantile regression via random forest.This is called “Quantile Control Method”(QCM).
Monte Carlos simulations show good coverage probability for the confidence intervals, which are robust to heteroskedasticity, autocorrelation, and model misspecification.
主讲人简介:
方诚,讲师。山东大学数量经济学博士。担任Economic Modelling匿名审稿人。主要研究领域为治理政策绩效评估、模拟应用研究和计量模型改进。参与国家社会科学基金项目2项,山东省省自然科学基金重点项目1项。以第一作者在《经济学(季刊)》等权威期刊发表论文。在校学术成果获山东大学优秀学术论文一等奖。
欢迎各位老师和同学踊跃参加!