Statistical Methods in Finance 2019

Dec 16 - 21, 2019


Censored Quantile Regression Models with a Cure Proportion

by Naveen Narisetty, University of Illinois at Urbana-Champaign

Quantile regression is a powerful and flexible tool that has found widespread use in financial applications. This is because the quantile regression model can provide a more robust and comprehensive description of the relationship between a financial outcome and covariates of interest. A new quantile regression model for censored data is proposed that permits a positive proportion of observations to become insusceptible to the event of interest. In contrast to prior proposals for estimation of censored quantile regression models, we propose a new "data augmentation" approach for estimation of the model. Our approach has computational advantages over earlier approaches and empirical studies demonstrate its superior statistical performance.