Fair Lending and Insurance: Disparate Impact Analysis

Disparate Impact in Insurance

Disparate impact occurs when a facially neutral rating variable disproportionately affects a protected class. In insurance, variables like credit score, education level, occupation, and ZIP code may serve as proxies for race, ethnicity, or income. Regulators and consumer advocates have increasingly scrutinized insurance pricing models for disparate impact, even when the models do not explicitly use protected class variables. Actuaries must balance the predictive accuracy of their models with fairness considerations, navigating a complex landscape of state regulations, federal oversight, and evolving social expectations.

Analytical Approaches

Actuaries use several techniques to assess disparate impact. Adverse impact ratios compare approval rates, pricing, or loss ratios across protected and non-protected groups. Regression analysis can test whether disparities persist after controlling for legitimate risk factors. Proxy discrimination testing evaluates whether removing a suspect variable changes the model's differential impact on protected classes. Bias mitigation techniques include constrained optimization (limiting the disparate impact while preserving predictive accuracy), variable exclusion, and post-hoc adjustments to model outputs. The challenge lies in defining appropriate fairness metrics and balancing competing objectives: actuarial accuracy, regulatory compliance, business viability, and social equity.