Conveying the Implications of Risk Selection and Risk Adjustment Strategies to Policy Makers

• Activity Number: 379
• Type: Luncheons
• Date/Time: Wednesday, August 6, 2003 : 12:30 PM to 1:50 PM
• Sponsor: Section on Health Policy Statistics
• Author(s): Harold Luft*+
• Companies: Institute for Health Policy Studies
• Keywords: risk adjustment ; nonrandom ; policy maker ; simulation ; graphical presentation ; Monte Carlo

Abstract

Whenever potential enrollees or patients can choose among health plans or provider groups, biased selection can occur through (1) dumping, in which people who have been high-cost in the past are avoided; (2) skimming, in which people who have been low-cost in the past are attracted, and (3) stinting, in which -- across the expenditure distribution -- people who are somewhat higher-cost are less likely to be enrolled than those who are somewhat lower-cost. Each strategy has different implications for the mix of enrollees in favorably and adversely selected plans, and for the performance of alternative approaches to risk adjustment. Policy researchers have traditionally focused on reporting the R2 values for the various risk adjustment models that they develop. We evaluate such models using a new simulation model (Nonrandom Risk Adjustment) that takes a Monte Carlo approach to replicate each of the three types of selection. With this model we can incorporate person-level reinsurance and flexibly examine various distributional aspects of the results (such as how often a risk adjusted payment is within 5% of observed expenditures), not just summary measures.