Health Plan Participation in the Medicare Advantage Market
Authors: Emily Adrion, Brad Herring
The payment and regulatory policies facing Medicare managed care plans have changed considerably over time, but rather than being evidence-based, these policy changes have often been politically motivated. Modifications to payment policies facing Medicare managed care plans, such as those in the 2010 Affordable Care Act (ACA), have generated fears of health insurance market instability. Many often predict that plans, when faced with lower payment rates, will pull out of the market causing disruptions in coverage and leaving Medicare beneficiaries without access to private plans. The purpose of this paper is to examine the empirical relationship between Medicare Advantage (MA) plan participation and MA payment rates, as well as the relationship between MA plan participation and other market characteristics.
Using Centers for Medicare and Medicaid Services (CMS) MA Contract, Enrollment, and Rate Calculation data files for years 2008 and 2011, along with other secondary data, we examine county-level MA plan participation using three approaches: cross-sectional logistic regressions for plan participation at a point in time; first-difference multinomial logit regressions for plan entry and exit; and first-difference OLS regressions for changes in county MA penetration. County benchmark payment rates and other market characteristics are key explanatory variables, along with interactions between the benchmark rate and plan profit status and plan type.
Our findings indicate that a 10.1 percentage point (or two standard deviation) reduction in the relative benchmark rate (defined as the ratio of the county benchmark rate to average per-capita costs under fee-for-service Medicare in a county) between 2008 and 2011 was associated with a 2 percentage point lower predicted probability of plan entry (i.e., 66% vs. 68%) and a 2 percentage point higher predicted probability of plan exit (i.e., 66% vs. 64%) compared to the predicted probability of entry and exit at the mean change in the relative benchmark rate of -2.5 percentage points. We also find that the relationship between the benchmark rate and plan participation is stronger among for-profit plans as compared to nonprofit plans and that plan participation is higher in areas with less-concentrated hospital markets and by plans already participating in non-MA markets in an area.
Our results uniformly suggest that there was a positive and statistically significant association between the relative benchmark rate and plan participation over the study period; these findings hold both across counties at a point in time and within counties over time. However, predicted probability calculations for market entry and exit from 2008 to 2011 show that changes in the relative benchmark rate were associated with relatively modest changes in plan participation.
MedPAC has consistently recommended that MA benchmarks be lowered to be on equal footing with fee-for-service Medicare. However, insurance industry representatives have countered that MA plans would not survive that level of payment cuts and would be forced to exit the market. We conclude that the magnitude of the relationship between benchmark rates and plan participation is not as strong as perhaps conventionally believed.