Fair Regression in Health Spending

Risk adjustment formulas predict spending in health insurance markets in order to provide fair benefits and health care coverage for all enrollees, regardless of their health status. Unfortunately, current risk adjustment formulas are known to undercompensate payments to health insurers for specific groups of enrollees (by underpredicting their spending). Much of the existing algorithmic fairness literature has focused on classifiers and binary outcomes. In this article, we expand concepts from the statistics, computer science, and health economics literature to develop new 'fairregression’ statistical machine learning methods for continuous outcomes in an effort to improve risk adjustment formulas for undercompensated groups. We additionally propose novel definitions of fairness for this setting. Our data application using the Truven MarketScan Commercial Claims and Encounters database demonstrates that alternative methods for risk adjustment formulas perform better than current techniques with respect to an undercompensated group, and that a suite of metrics is necessary in order to evaluate the formulas more fully.