Speaker: Dr. 杨珅珅 (Shenshen Yang)

Title: A GMM Approach Towards Multi-stage Non-monotone Missingness

Schedule: April 10, Wed 2:00-3:30 PM

Location: 诚明楼（Chengming Hall） RM315

Introduction: 杨珅珅，天津大学马寅初经济学院助理教授，德克萨斯大学奥斯汀分校博士。研究领域包括计量经济学、劳动经济学、公共经济学。研究成果发表在国际计量学顶刊 Journal of Econometrics。详情请见个人网站 http://mysoe.tju.edu.cn/info/1242/3116.htm

Abstract: This study investigates the prevalent issue of non-monotone missingness in sequentially collected data. A new MAR-type assumption is introduced for identification purpose, upon which a two-step asymmetric AIPW-GMM estimator is constructed. This proposed estimator is double robust and meets the semiparametric efficiency bound under certain conditions. This method is applied to the Oregon Health Insurance Experiment, dealing with nonmonotone missingness on both the treatment and outcome variables. We highlight significant effects of enrollment in the Oregon Health Plan in enhancing health-related outcomes and reducing out-of-pocket medical care expenses. Evidence suggests that excluding incomplete data results in downward biases for selected outcome variables. Furthermore, the estimator introduced in this paper reduces the standard error between 6-24% for the estimated effects of the Oregon Health Plan, demonstrating its efficiency and robustness.