Optimal pricing of employer-based Health Plans

描述(由申请人提供)：雇主在管理其健康计划选项时面临的最大挑战之一是确定员工保费缴费结构。大多数雇主都受到成本控制的驱动，希望为员工提供一种激励，让他们参加高效的计划。然而，定价方案通常忽略了这样一个事实，即由于患者选择的影响，效率不能通过简单地比较不同计划的人均成本来确定。例如，选择的影响可能会导致封闭式医疗保健组织看起来效率不成比例，因为健康的人会转向这种低成本的选择。本研究将探讨定价、员工选择和由此产生的平均成本之间的相互关系，试图将选择的影响从计划效率对平均成本的影响中分离出来。具体地说，申请人将： 1.建立了一个基于效用的选择模型来预测患者对方案的选择，给出了贡献结构； 2.开发成本模型，探索患者选择模式如何影响计划的平均成本， 允许选择的影响与其他计划成本动因分开； 3.模拟不同定价理念对自保和全保的影响 场景。 这些模型将基于一家拥有四个计划选项的大型雇主的三年登记数据和两年申领数据。详细的索赔数据允许使用约翰·霍普金斯调整后的临床小组(ACG)病例组合软件来开发每个人的健康状况的衡量标准，作为建模中的协变量。基于Gibbs抽样的贝叶斯推断将被用来估计健康计划选举的多项概率选择模型。申请人将评估两种分级成本模型，这两种模型考虑了寻求护理的选择效应，并采用随机效应来捕捉家庭内部相关性的影响。使用这些模型作为工具，将在不同的雇主定价理念下探索计划提供的稳定性：保费缴费百分比、固定美元缴费和根据计划的健康风险组合调整的固定美元缴费。在完全投保的情况下，问题扩展到解决保险公司的纳什均衡问题，因为保险公司在知道雇主的缴费策略的情况下设定了他们的保费水平。

DESCRIPTION (provided by the applicant): One of the biggest challenges for employers in the management of their health plan options is determining the employee premium contribution structure. Most employers are driven by cost control, and want to create an incentive for employees to enroll in efficient plans. However, pricing schemes typically ignore the fact that efficiency cannot be determined by a simple comparison of average per-person costs across plans, because of the influence of patient selection. For example, the impact of selection may cause a closed-panel HMO to appear disproportionately efficient, as healthy people migrate to this low-cost option. This research will explore the interrelationships between pricing, employee choice and the resulting average costs, in an effort to separate the selection impact from the influence of plan efficiency on average costs. Specifically, the applicant will: 1. Develop a utility-based choice model to predict the patient's choice of plan, given contribution structure; 2. Develop a cost model to explore how patterns of patient choice influence average cost by plan, allowing the influence of selection to be separated from other plan cost drivers; 3. Model the impact of different pricing philosophies in both self-insured and fully-insured scenarios. These models will be based on three years of enrollment data and two years of claims data from a large employer with four plan options. Detailed claims data allows the use of the Johns Hopkins Adjusted Clinical Groups (ACG) Case-Mix software to develop measures of the health status of each individual, to be used as a covariate in the modeling. Bayesian inference with Gibbs sampling will be used to estimate a multinomial probit choice model for health plan elections. The applicant will evaluate two hierarchical cost models that take into account the selection effect of seeking care, with random effects to capture the influence of correlation within families. Using these models as tools, the stability of the plan offerings will be explored under various employer pricing philosophies: percent of premium contribution, fixed dollar contribution, and fixed dollar contribution adjusted for the plans' health risk mix. In the fully-insured scenario, the problem extends to solving the insurers' Nash equilibrium problem, as they set their premium levels, given knowledge of the employer's contribution strategy.