A Review of Gender Bias Mitigation in Credit Scoring Models

Literature reviews on metrics to quantify biases and techniques to mitigate them have been made in the past. However, none of these investigations have centered on the application of these methods to the specific fields of credit scoring, or gender bias. In this article, we present a literature review on three databases (ACM Digital Library, IEEE Xplore, Springer Link) to find out what types of metrics and bias mitigation techniques have been used in credit scoring models to quantify or reduce gender bias. Only published experiments or case studies in English were included. We found that most articles are centered on pre-processing techniques for bias mitigation, using mainly causal fairness measuring methods. No article was found that focused specifically on gender issues; they all presented gender bias just as a possible case of application of the presented techniques. We conclude with some findings regarding possible research paths: mainly doing more case studies, considering collections scores, and further consideration of scores as rankings.