Association for Public Policy Analysis & Management

After the 2018 election, several states appear poised to expand Medicaid under the Affordable Care Act, while other states are considering partial expansions (to 100% of the federal poverty level [FPL]) in order to maintain a larger role for private insurance. Meanwhile, several 2020 Presidential candidates have proposed replacing private insurance with a single public payer model. In this context, understanding the tradeoffs between these coverage types is critical. We use a regression discontinuity design (RD) and a unique dataset combining income information and all-payer claims to compare public vs. private insurance coverage.

Merging income data from Colorado’s Medicaid and Marketplace agencies with the state’s All Payer Claims Database from 2014, we examined utilization, out-of-pocket spending, and total costs using RD methods to analyze the eligibility discontinuity that occurs at 138% of FPL. Our sample contained non-pregnant individuals ages 19-64, with incomes between 75% and 400% of FPL (N=223,147). Comparing individuals just below versus just above the eligibility threshold of 138% of FPL, we assess the impact of coverage type of these health-care related outcomes.

Our preliminary findings are threefold. First, income below the Medicaid eligibility threshold led to 7.0 months of additional Medicaid coverage per year, while those just above the threshold experienced 6.3 months more of Marketplace coverage. This implies that Medicaid eligibility, with less cost-sharing and no premiums, led to 10% more months of coverage among enrollees.

Second, utilization patterns differed significantly. The RD estimates indicate that those eligible for Medicaid were 4 percentage-points more likely to visit the Emergency Department each year (sample mean 25%) than those eligible for Marketplace coverage, but were 4 percentage-points less likely to have an outpatient visit (sample mean 87%). Medicaid resulted in 0.18 additional ED visits per person and 0.19 fewer outpatient visits. Medicaid also led to a 3 percentage-point decrease in the likelihood of a prescription fill (sample mean 59%) and 0.86 fewer prescription fills per person (sample mean 8.6). We found no statistically significant differences in hospitalization rates.

Third, costs were significantly higher in Marketplace coverage, despite the slight reduction in months covered. Annual spending per person was $741 higher in Marketplace coverage than Medicaid, representing a 36% increase compared to the sample mean ($2078). After standardizing prices to Medicaid rates, costs were similar in the two groups, suggesting that price differentials were responsible the large cost differences. Out-of-pocket spending (not including premiums) was even more disparate. Marketplace beneficiaries spent $104 more per person than Medicaid enrollees, more than a fivefold increase compared to the Medicaid average ($23). Results were generally robust to parametric and local linear RD models, and variable bandwidths.

Overall, our results indicate that there are meaningful differences in the health care experiences of low-income adults in public vs. private coverage. Higher ED and lower office visit rates in Medicaid may reflect greater difficulty in accessing outpatient providers, as well as lower cost-sharing barriers to ED care in Medicaid than private insurance. Meanwhile, Medicaid coverage is substantially less costly both to enrollees and to society.