Duke 2008 | ASHE | Equity and Efficiency in Health and Healthcare

Consumer Driven Health Plans (CDHPs) have grown in popularity among large employers since their introduction in 2001. Approximately 3 million people are enrolled in CDHPs, through employer group coverage. Established health insurers such as Blue Cross Blue Shield, Aetna, UnitedHealth Group and Cigna are offering CDHPs in addition to their traditional insurance products. Several early studies have focused on the financial impact of these plan designs on the cost to employers, but none have examined the difference in the welfare effects to consumers between CDHPs and alternative forms of health insurance.

Research Questions We examine the welfare impact of health insurance innovation. Our analysis is in the spirit of Petrin (2002). The introduction of CDHPs is a new health insurance product and we measure the impact of its introduction on the welfare of health insurance consumers. Health insurance is a differentiated product and in differentiated product markets entry of new products can reduce welfare (Mankiw and Winston, 1986). The welfare impact will depend upon any efficiencies cost of providing health insurance and the value that consumers place upon have access to new insurance structures. Furthermore, the welfare of the introduction of the CDHP will not depend only upon the distribution of the value of the product across consumers and its marginal costs but also on its impact on the equilibrium pricing. In the presence of adverse selection the welfare outcomes of new health insurance product introductions are ambiguous (Cutler and Reber, 1998) as new products may exacerbate adverse selection and may lead to pricing distortions. The authors address the following two questions in their paper:

How did the introduction of CDHPs affect consumer surplus? Who gained and lost welfare from its introduction? What are the total welfare consequences of the introduction of CDHP? This analysis will estimate the cost functions of CDHPs and the impact of its introduction on the pricing of health insurance plans.

Methods We will use a structural equation approach to model the impact of CDHPs. We estimate the demand for health insurance using a multinomial probit choice model that explicitly includes the expected out-of-pocket expenditures of the enrollee. This model allows for persistent consumer heterogeneity and flexible substitution patterns between health insurance products. Our model will explicitly incorporate the possibility that there is both selection on observables and un-observables. Because of the structure of our choice model, the consumer calculations will not be driven by many of the unappealing features of a logit or nested- logit specification. We will also estimate is the probability of the use and the total cost of any medical care use, including prescription drugs conditional on plan choice. Independent variables include patient illness burden prior to the CDHP choice year, age, gender, zip code associated regional attributes, wages, number of dependents, marital status and job classification. Finally, we analyze the impact of the introduction of CDHP on premiums both in the employer context and in a private health insurance market. The latter analysis will be performed using simulations methods.

The computation complexity of our approach necessitates a Bayesian approach. We estimate cost and choice parameters simultaneously using an MCMC simulator. Specifically, posterior distributions are approximate using Gibbs sampling.

The data used for this analysis is from a large employer with national markets and substantial operations in 22 US States. The total population examined over a three year period of time 120,000 covered lives. There are up to 6 separate health plan choices available to each employee. In every market, there are at least two types of CDHP available. The data used spans services reimbursed between calendar years 2002 through 2005.

Results Early results used in our analysis have shown there to be little adverse selection with respect to chronic illness. If anything, CDHPs attracted less favorable selection. Correspondingly, CDHP costs increase substantially, following case-mix adjustment and constitute the largest percentage increase over a two period of time in comparison to alternative health insurance designs. Where the CDHP achieves the lowest impact in expenditure is in pharmaceuticals. With respect to out of pocket expenditures, CDHPs appear to be neutral in their impact based on descriptive analyses. However, we have not yet employed a full range of consumer welfare variables in a system of equation approach. These results will be shared at the meetings.

The finding of this analysis will be the first empirical evaluation of this topic. These findings will provide evidence of whether CDHP choice has any welfare decreasing effects. As these plans grow in popularity and become a subject of public policy debate as Health Savings Accounts (HSAs) become politicized, our results will inform the question of the long term viability of this new insurance product.