Academy Recognizes Anti-Discrimination Paper With Research Award

By Julia Goodwin
Senior Research Analyst

The Academy presented its third annual Award for Research to Xi Xin for his paper, Anti-discrimination Insurance Pricing: Regulations, Fairness Criteria, and Models at the 2025 Actuarial Research Conference. The award recognizes an early-career scholar whose work provides an actuarial perspective on a public policy issue of interest to U.S. actuaries and policymakers. Grace Lattyak, chairperson of the Academy's Research Committee, presented the award to Xin at the conference which was hosted by York University in Toronto, July 29-Aug. 1.

This year's theme was "Bias in assessing financial risk: Origins, detection, mitigation of biases that might affect actuarial assessments in insurance, retirement planning, and/or financial risk assessment." The theme speaks to the Academy's research and policy focus, as well as that of the National Association of Insurance Commissioners, on disparate treatment, discrimination, and bias-and their mitigation-in insurance or financial risk, especially amid the rising use of machine learning and artificial intelligence (AI) models in insurance.

Xin's paper, co-authored with his advisor Fei Huang, discusses how fairness concepts from economics, law, and machine learning apply to insurance pricing regulations, particularly around proxy discrimination. For example, ZIP codes or credit scores are often used as proxies to uncover protected characteristics like race. Running parallel to this issue of bias is the influx of fairness criteria in the burgeoning area of machine learning and AI.

The paper succinctly outlines definitions of bias and fairness criteria, matches fairness criteria with potential and existing anti-discrimination regulations, and implements these models into a series of existing and newly proposed antidiscrimination insurance pricing models.

The Award for Research Work Group decided on Xin because the paper provided a thorough assessment of bias and fairness, merging the concepts into useful theoretical models. Frank Todisco, a Work Group member and incoming president-elect of the Academy, noted that "one of the reasons for giving Xin the edge was his use of real data and his interesting empirical findings … [his research covered] both concepts and laws and regulations in a variety of jurisdictions and lines of business over time. In this regard, Xin's paper will be especially useful to me and my staff in our actuarial work at GAO [Government Accountability Office]." Xin, who is a Ph.D. candidate in the School of Risk and Actuarial Studies at the University of New South Wales in Sydney, Australia, accepted the award and $7,500 prize via Zoom.

Preceding the award's presentation, the Academy sponsored a panel session during the conference, which included Xin and two of the Academy research award finalists-Olivier Coté, a Ph.D. candidate in the School of Actuarial Science at Université Laval and Tianhe Zhang, a Ph.D. candidate in the Risk Management and Insurance Department at the University of Wisconsin, Madison. The three presented their work and engaged in an intriguing and lively discussion, moderated by Lattyak.

Coté's work focused on the use of causal graphs to capture direct and indirect discrimination based on prohibited characteristics (e.g., race, age, gender) and their proxies, linking discrimination models with fairness criteria to illustrate the interplay between fairness and discrimination. Zhang's work provided a theoretical model for how insurance companies can create workflows between themselves and trusted third parties, where the third party would securely store "noisy" sensitive attributes and combine that data with the insurer's non-sensitive data. The third party would subsequently build a model that assesses risk without direct access to sensitive attributes, providing insurance companies with a pricing mechanism that would correct for discrimination based on the protected characteristics.

The discussion that followed focused on the tension between the differing concepts of fairness. While actuarial fairness focuses on an insurance company's ability to match premiums with risks, it may run contrary to social fairness, whereby insurance companies may be required by law to assist specified groups. The panelists highlighted the necessity for transparency in insurance pricing models and how regulators and policymakers can best proceed to reduce or eliminate bias in insurance pricing.

This year's Award for Research underscores the connection between public policy and research. It not only examines the value in advancing an understanding of bias in insurance pricing, but also offers an opportunity and potential springboard for future Academy work. We know that several of our practice councils have been actively engaged in conversations with their volunteers and external audiences on practical and theoretical applications of actuarial expertise when identifying or addressing different forms of bias.

The value of the Academy's annual Research award is to help draw additional interest and spark conversations on future work that can be used to fulfill the Academy's mission. For those interested in learning more about the winning research, as well as the finalists, stay tuned for a webinar in early 2026 that will feature Xi Xin. For more on the Academy's engagement in research, check out our website and keep an eye out for the 2026 announcement of the theme for next year's Academy Annual Award for Research.