A COMPARATIVE ANALYSIS OF ACTUARIAL-BASED CREDIT RISK MODELS FOR LOW-INCOME POLICYHOLDERS: CHALLENGES AND THEORETICAL SOLUTIONS

This paper investigates the limitations of actuarial-based credit risk models for low-income policyholders and explores solutions to reduce structural bias. Traditional models rely heavily on credit histories and proxy variables, which systematically disadvantage individuals with limited financial records. A comparative review highlights both the predictive shortcomings and fairness challenges of current approaches. To address these issues, a dual-model pilot framework is proposed, contrasting conventional generalized linear models with enhanced models incorporating alternative data such as utility payments, rental histories, and mobile money transactions. Evaluation metrics include predictive accuracy, fairness indicators such as disparate impact ratios, and reclassification outcomes for low-income applicants. The study demonstrates that integrating alternative data can strengthen predictive reliability while reducing inequities in access to insurance and credit. Policy implications underscore the need for transparency, fairness-aware actuarial standards, and regulatory oversight to ensure credit risk models balance solvency with social responsibility. Keywords: Actuarial Models, Credit Risk, Low-Income Policyholders, Fairness in Insurance, Alternative Data