Identifying Upcoding in Markets with Adverse Selection: An Application to Medicare

Risk-adjustment is commonly used in health insurance markets to deal with problems of adverse selection and cream skimming by compensating health plans for insuring consumers whose diagnoses are more costly to treat. However, in all real world risk adjustment systems, it is the insurers themselves who report the diagnoses that determine risk scores. This creates incentives to "upcode" enrollees to extract higher payments. The existence and extent of upcoding is largely unknown because in most data, upcoding is observationally equivalent to adverse selection, in which consumers systematically sort across health plans on the basis of their health risk.

We model upcoding in the presence of adverse selection. Our model delivers a novel strategy for empirically separating upcoding from selection in aggregate, market-level data. The intuition behind our approach is that if the same individual generates a higher risk score in one plan than another, then as the market share of the more intensively coded plan increases, the overall level of reported risk in that market should increase. Such a pattern can not be due to selection, because all enrollees are included in the calculation of the market-level risk score, regardless of the plan they selected into. We apply this strategy to analyze upcoding by Medicare Advantage plans. The results show that enrollees in Medicare Advantage plans generate 5% higher risk scores than what the same enrollees would generate under Traditional Medicare. Absent a coding inflation correction, this implies a distortion in seniors’ choice between Medicare Advantage and Traditional Medicare, and excess payments to Medicare Advantage of around $6 billion annually. Our findings also have implications for the geographic distribution of Medicare's resources, as Medicare Advantage penetration rates vary widely across localities.