Premium and Cross-Premium Elasticity of Medicare Health Plan Choices
Data and Methods: The primary data source is the Medicare Current Beneficiary Survey (MCBS). The MCBS is a representative sample of the Medicare population with information on beneficiary plan choices, spending and beneficiary characteristics. Data are drawn from the 2011 survey. These are combined with data from Medicare Compare on MA plan characteristics and insurer data on Medigap premiums. Sample exclusions include having an employer Medicare supplement, living in a facility, and Medicaid eligibility due to differing choice sets for those individuals. The total sample size was 4,237. The unit of analysis is the individual beneficiary.
Plan choices are estimated using nested logit with nests defined by MA and FFS. Six different choices are: FFS, FFS+Medigap, FFS+Part D, FFS+Medigap+Part D, MA alone and MA+Part D. This is the first time that all six options have been included in a Medicare plan choice model. Key plan characteristics are the premium and benefit design. Premium is the total cost of the plan, including MA, Part D and Medigap premiums. Premiums and (for some options) benefits vary geographically depending many factors, including MA payment levels, FFS beneficiary characteristics and state Medigap premium regulations. Individual characteristics include socio-demographic and health variables such as age, income, education, chronic illnesses and self-rated health. Individual choices enter the model both at the cross-sector (FFS / MA) choice and via interaction terms for the within-sector choice.
Preliminary Results: The most popular choices were MA+Part D (n=1,817) and FFS+Medigap+Part D (n=1,161), followed by FFS+Medigap, FFS only, FFS+Part D (no Medigap) and finally MA without Part D. Preliminary results show that the premium was negative (b= -0.0019) and statistically significant (p=.001). The average price elasticity of demand is equal to -.059. Plans with lower cost sharing were also preferred. The effect of other variables (age, gender, income, health status) varied by sector. Controlling for other factors, chronic illnesses increased the probability of selecting FFS+Medigap+Part D, followed by FFS+Medigap. Income had a similar pattern – increased the probability of the FFS choices with Medigap, but not the different Part D options.