When evaluating outcomes and potential bias, statistical methods for imputing or modeling race and ethnicity can be helpful tools. While various approaches have been used to address missing race and ethnicity data over the years, imputation methods have evolved significantly and not all are not widely known or understood in the actuarial profession. We will describe a range of techniques for developing probabilistic estimates or predictions of individual race and/or ethnicity.
In addition, this session will explore a novel assessment and management framework around unfair discrimination, with specific application for the potential impact of race in a rating model. This methodology is being used to understand potential unfair discrimination and respond to the current and future regulatory actions around variables.
Learning Objectives:
Understand different imputation methods for race and ethnicity.
Learn about the main packages for imputing race and ethnicity.
Understand a methodology that can be used to assess the impact of race.
Understand various methods that can be used for analyses of and testing for unfair discrimination.
