The overarching goal of this project is to evaluate the extent of switching costs in Medicare health plans to understand whether they are large enough to undermine the ability of competitive markets to function in a premium support context. To do this, we will analyze whether Medicare beneficiaries are sufficiently responsive to changes in health plan premiums and benefits for "premium support" style Medicare reforms to be a viable strategy to control Medicare spending. The Medicare program faces substantial financial challenges over the next several decades. The Medicare Trustees project that the Part A trust fund will become insolvent by 2025 and that the entire program is on an unsustainable spending trajectory. Although there are many different proposals for how to deal with the impending spending crisis, most falls into one of two groups: care redesign and premium support. Care redesign seeks to preserve the basic structure of the existing Fee-For-Service (FFS) model while introducing incentives to provide care more efficiently. The second idea - premium support - has health plans submit "bids" to the government for a basic benefit package. The government uses these bids to set its contribution to premiums. Plans that provide better value are rewarded with high market share, while those that do not are penalized by losing market share. This is the essential logic behind the structure of Medicare Part D prescription drug coverage, a program that has achieved high beneficiary satisfaction and program costs substantially below projections. However, whether this Part D result would translate to a premium support model that included broader coverage is unclear. Although there have been substantial efforts to evaluate the impact of care redesign in Medicare - CMS is currently funding numerous evaluations of care redesigns, such as bundled payment initiatives and Accountable Care Organizations - there has been relatively little effort to evaluate the viability of premium suppor models, despite its prominence as a possible reform option. The project draws data from the Medicare Current Beneficiary Survey, a representative longitudinal dataset. Analytically, we will estimate the magnitude of switching costs using a mixed logit model, observing plan choices over time. We will explicitly model the impact of plan "loyalty" on plan choices, and will allow "loyalty" to be a random coefficient. The random coefficient will allow us to identify not only the mean value of loyalty, but also to identify factors which predict loyalty.