Venue: The Fuqua School of Business, Duke University, 1 Towerview Drive, Durham, NC 27708-0120
Presentation
Medicare Beneficiaries' Choice of Medicare Health Plans - A Choice-with-Screening Model
Recently, the Medicare reform has focused on expanding the variety and the availability of Medicare health pans, which complicates the plan choice for the elderly. The neo-classical theories emphasize that the elderly ill benefit from the expanded choices. However, evidence from behavioral economics finds that the consumers may use screens to simplify their choice situation. This paper investigates the screening behavior of the elderly in the choice of Medicare health plans.
A choice-with-screening model is proposed, where the decision process of the elderly follows two stages. In the first stage, the elderly screen all the available health plans for future consideration; in the second stage, they make the final choice among the plans that passed the screening. It is assumed that a health plan can pass the screening if and only if all its attributes can satisfy the individual's requirement. Moreover, to control the individual's interest in status quo, the default Medicare Fee-For- Service plan is assumed to always pass the screening. After the screening, the plan choice is analyzed by the random coefficient multinomial probit (RCMNP) model. The individual requirement for attributes in the screening and the individual preference for attributes after the screening depend on the demographics.
The choice-with-screening model is estimated in the Bayesian framework facilitated with Markov-chain Monte Carlo methods. The choice information and demographics of the elderly are obtained from the Medicare Current Beneficiary Survey and the information of Medicare health plans is from the Medicare Health Plan Compare dataset. The results show that the elderly are likely to screen Medicare health plans according to premium, prescription drug coverage and vision service coverage. Compared with the conventional RCMNP model, the choice-with-screening model fits the data better and can capture certain nonlinearities in the effects of plan attribute.