Title: Abadie's Kappa and weighting estimators of the local average treatment effect
Authors: Tymon Sloczynski - Brandeis University (United States) [presenting]
Derya Uysal - University of Munich (Germany)
Jeffrey Wooldridge - Michigan State University (United States)
Abstract: The finite sample properties of various weighting estimators of the local average treatment effect (LATE), several of which are based on kappa theorem, are studied. We argue that one of the Abadie estimators, which we show is normalized, is likely to dominate the other in many contexts. A notable exception is in settings with one-sided noncompliance, where certain unnormalized estimators have the advantage of being based on a denominator that is bounded away from zero. We use a simulation study and an empirical application to illustrate our findings. While the performance of kappa-based estimators appears to be context-dependent, the normalized estimator performs very well in all settings under consideration.