Health Insurance Transition Patterns Before and After the ACA: Evidence from Three States

Health insurance coverage in the US is often subject to instability due to changes in employment or family income. Although the creation of the Exchanges and the expansion of Medicaid under the Affordable Care Act (ACA) have expanded opportunities for coverage and reduced uninsurance rates, research suggests that these new coverage options and changes in the composition of the insured could result in an increased rate of transitions across coverage sources, and between insurance and uninsurance.

Evidence on patterns of health insurance transitions under the ACA remains limited, in part because of limitations in widely used survey and administrative datasets. Survey datasets are subject to measurement error due to respondent misreporting, furthermore very large sample sizes are necessary to obtain precise estimates of transition rates. Typical administrative data sources on health insurance enrollment may be inappropriate for studying transitions between public and private coverage sources and may consist of unrepresentative samples that cannot support inference about population-level transition behavior. State All-Payer Claims Databases hold considerable potential to overcome these data limitations, since they contain panel data on the universe of enrollees in both commercially insured and publicly funded health plans.

In this paper, we use All Payer Claims Databases from three Medicaid expansion states (Arkansas, Colorado, and Rhode Island) to estimate insurance transition rates, with an emphasis on transitions between Medicaid and private insurance. Our data span the implementation, in 2014, of the major ACA coverage expansions, enabling us to compare transition rates across sources of coverage before and after the ACA took effect. By using APCD data from multiple states, we are also able to go beyond existing analyses of transition rates using nationwide survey data and examine whether different state approaches to Medicaid expansion affect the stability of coverage or transition patterns for those enrolled in new coverage sources. Arkansas, Colorado, and Rhode Island each used different approaches to providing coverage to the Medicaid expansion population: Rhode Island enrolled the expansion population in existing Medicaid Managed Care programs, Colorado enrolled the expansion population in fee-for-service Medicaid provided through the state Medicaid agency, while Arkansas adopted a unique premium support model (known as the "private option") whereby Medicaid-eligible individuals are enrolled in private health coverage provided through the state's insurance Exchange.

While analysis of Rhode Island and Arkansas data is in progress, our results from Colorado indicate three main findings. First, overall, stability in year-on-year insurance coverage is the norm; the majority of individuals remain covered by the same source of insurance from year to year. Second, transitions into Medicaid become more common after 2014, with individuals previously covered by ESI and Individual (non-Exchange) insurance transitioning to Medicaid at much higher rates. Third, the stability of Medicaid coverage increases in 2014 and later years, with enrollees are much more likely to remain on Medicaid after the expansion of the ACA. Comparison of these patterns across all three states will provide new evidence on whether different models for providing Medicaid coverage are associated with meaningfully different coverage transition patterns.