Title: Adaptive Rank Inference in Semiparametric Multinomial Response Models
Speaker：Shakeeb Khan
Institution: Duke University
Time: June 20, 2:00-3:30pm
Location: RM614, Fanhai Building
Abstract：We consider estimation and inference on regression coeffcients in semi parametric multinomial response models. Our approach is based on a localized rank objective function, loosely analogous to that used in Abrevaya et al. (2010), which we show achieves sharp identification of the identi_ed region. This is in contrast to existing procedures in the literature such as Ahn et al.(2014), which although providing con-sistent estimators when conditions for point identification are satisfied, do not yield sharp set estimates when they are not. A leading case when point identification is not satis_ed is when all covariates are discrete, as happens often in empirical settings. In these settings we show that the aforementioned procedures sometimes only estimate the trivial set. In contrast, our procedure is adaptive in the sense that it provides an estimator of the sharp set when point identification does not hold, and a root-n consistent point estimator when it does. Furthermore, we show our rank procedure readily extends to panel data settings for inference in models with fixed effects. This includes dynamic panel models with lagged dependent variables as covariates. A simulation study establishes adequate finite sample properties of our new procedures.
返回顶部
