A0695
Title: Bayesian model averaging in causal instrumental variable models
Authors: Gregor Steiner - University of Warwick (United Kingdom) [presenting]
Mark Steel - University of Warwick (United Kingdom)
Abstract: Instrumental variables are a popular tool to infer causal effects under unobserved confounding, but choosing suitable instruments is challenging in practice. The aim is to propose gIVBMA, a Bayesian model averaging procedure that addresses this challenge by averaging across different sets of instrumental variables and covariates in a structural equation model. The approach extends previous work through a scale-invariant prior structure and accommodates non-Gaussian outcomes and treatments, offering greater flexibility than existing methods. The computational strategy uses conditional Bayes factors to update models separately for the outcome and treatments. It is proven that this model selection procedure is consistent. By explicitly accounting for model uncertainty, gIVBMA allows instruments and covariates to switch roles and provides robustness against invalid instruments. In simulation experiments, gIVBMA outperforms current state-of-the-art methods. Its usefulness is demonstrated in an empirical application, estimating the causal effect of education on income.
