A0868
Title: Type II Tobit sample selection models with Bayesian additive regression trees
Authors: Eoghan O Neill - Erasmus University Rotterdam (Netherlands) [presenting]
Abstract: The purpose is to introduce type II Tobit Bayesian additive regression trees (TOBART-2). BART can produce accurate individual-specific treatment effect estimates. However, in practice, estimates are often biased by sample selection. The type II Tobit sample selection model is extended to account for nonlinearities and model uncertainty by including sums of trees in both the selection and outcome equations. A Dirichlet process mixture distribution for the error term allows for departure from the assumption of bivariate normally distributed errors. Soft trees and a Dirichlet prior to splitting probabilities improve the modelling of smooth and sparse data-generating processes. A simulation study and an application to the RAND Health Insurance Experiment data set are included.
