A0761
Title: The encouraged LATE: Point identified treatment effects without the exclusion restriction
Authors: Shuo Jiang - Xiamen University (China) [presenting]
Jinyoung Choi - Hankuk University of Foreign Studies (Austria)
Arthur Lewbel - Boston College (United States)
Abstract: Consider treatment effect estimation given a binary instrument that violates the standard exclusion restriction. The aim is to define and provide conditions (some testable) under which an encouraged local average treatment effect(E-LATE) is point identified. This E-LATE equals LATE when exclusion holds. A spectrum of identification results is provided under different scenarios. Under additional restrictions, a simple root-N consistent, asymptotically normal semiparametric estimator is provided. The estimator is applied to find the E-LATE of incumbency on election outcomes using the randomized order of candidates names on the ballot as an instrument without imposing the exclusion restriction.
