Weak Causality of a Generalized Wald Estimand
◈ 주 제: Weak Causality of a Generalized Wald Estimand
◈ 발표자: 신희원 (석박통합과정)
◈ 일 시: 2025년 5월 8일 목요일 11:30~13:00
◈ 장 소: 우석경제관(223동) 504호
◈ 주 관: 경제학부, 경제연구소 한국경제혁신센터, SSK, BK21
참여를 원하는 학생은 아래의 링크에서 신청 부탁드립니다.
https://forms.gle/6GJASFtG7XKr1ECu9
Abstract
In this seminar, I discuss weak causality of a generalized Wald estimand under a set of weaker assumptions than the monotonicity condition of Imbens and Angrist (1994). In the first part, I introduce the notions of weak causality, partial monotonicity and limited monotonicity, along with related results from the instrumental variable-unobserved heterogeneous treatment effects (IV-UHTE) literature. In the second part, I propose a generalized Wald estimand that is weakly causal under milder conditions, referred to as local monotonicity and first-stage scalar partial monotonicity. Lastly, I replicate Stephens and Yang (2014)’s estimation of returns to education in the U.S. The key implications are: (1) there exist weakly causal estimands that are easy to compute under less restrictive behavioral assumptions than Imbens and Angrist (1994)’s monotonicity and (2) partial monotonicity proposed by Mogstad, Torgovitsky and Walters (2021) is still too strong for weak causality in that under this assumption, a multitude of weakly causal estimands can be identified.