A0372
Title: Quasi-Bayesian estimation and inference with control functions
Authors: Ruixuan Liu - Chinese University of Hong Kong (Hong Kong) [presenting]
Abstract: A quasi-Bayesian method is explored that integrates frequentist estimation in the first stage with Bayesian inference in the second stage, motivated by structural discrete choice models using control function methodology to address endogeneity bias. In the first stage, a frequentist nonparametric approach estimates the control function, while the second stage employs a Bayesian approach to manage complex likelihood functions associated with the structural equation. The asymptotic properties of the quasi-posterior distributions are analyzed from the second stage, demonstrating that the resulting quasi-Bayesian credible set lacks the desired coverage in large samples. However, the quasi-Bayesian point estimator remains consistent and asymptotically equivalent to a frequentist two-stage estimator.
