ADVERSE SELECTION AND RISK RATING IN INSURANCE MARKETS

这项研究将使用1985-89年的人员数据和医疗保健索赔， 大型雇主开发一种测量和补偿 在相互竞争的健康保险计划中进行逆向选择。一个六方程 最大似然模型，用于预测医疗服务利用率， 将使用索赔数据作为因变量来估计支出 和人员数据作为独立变量。诊断信息 将用于识别特别高的风险 最有可能成为风险选择策略对象的个人; a 将开发各种特殊技术来调整雇主的风险 为这些人捐款。两组模型参数将 估计，一个是根据参加者在 一个是雇主的按服务收费计划，另一个是基于年度支出 对于参加预付费团体实践的人健康维护组织 (HMO)拥有很大的市场份额。FFS参数在 HMO参加者和FFS计划参加者的HMO参数将 根据收益产生两个可供选择的偏倚选择度量 这两种选择的做法的特点包和风格 布局预计支出也将产生的参加者在 雇主提供的几个较小的HMO来衡量逆向选择 在竞争的健康维护组织中。我们将分析预测和实际支出， 从FFS计划转换到HMO或从HMO转换到HMO的个人 另一个，并将其与预测和实际支出进行比较， 连续参加特定计划的个人。

This study will use 1985-89 personnel data and medical care claims from a large employer to develop a method for measuring and compensating for adverse selection among competing health insurance plans. A six equation maximum likelihood model for predicting medical care utilization and expenditures will be estimated using the claims data as dependent variables and the personnel data as independent variables. Diagnostic information from the claims data will be used to identify particularly high risk individuals most likely to be the objects of risk selection strategies; a variety of special techniques will be developed to risk-adjust employer contributions for these individuals. Two sets of model parameters will be estimated, one based on annual allowed charges incurred by enrollees in the employer's fee-for-service (FFS) plan and one based on annual expenditures for enrollees in a prepaid group practice Health Maintenance Organization (HMO) with a large market share. Application of the FFS parameters to the HMO enrollees and of the HMO parameters to the FFS plan enrollees will generate two alternative measures of biased selection based on the benefit packages and styles of practice characteristic of the two alternative plans. Predicted expenditures will also be generated for enrollees in the several smaller HMOs offered by the employer to measure adverse selection among competing HMOs. We will analyze predicted and actual expenditures for individuals switching from the FFS plan to an HMO or from one HMO to another and compare these to predicted and actual expenditures for individuals continuously enrolled in particular plans.