A0428
Title: Difference in differences with latent group structures
Authors: Hiroyuki Kasahara - University of British Columbia (Canada) [presenting]
Young Ahn - University of Pennsylvania (United States)
Abstract: The identification of average treatment effects on the treated (ATT) is examined within latent group structures, where the distribution of potential outcomes depends on latent types. A scenario is explored in which parallel trends are maintained when conditioned on latent types but may not hold in aggregate, resulting in an inconsistent standard difference-in-difference estimator. It is demonstrated that the latent group-specific ATT (LGATT) can be identified when parallel trend assumptions and other regularity conditions are met for latent types. An estimator for the LGATT is proposed that minimizes a weighted least squares criterion function, using weights derived from the estimated posterior probabilities of each latent type.
