Who Gained Insurance Coverage in 2014, the First Year of Full ACA Implementation?

Abstract The most significant pieces of the Affordable Care Act (exchanges, subsidies, Medicaid expansion, and individual mandate), implemented in 2014, were associated with sizable gains in coverage nationally that were divided equally between gains in Medicaid and private coverage. These national trends mask heterogeneity in gains by state Medicaid expansion status, age, income level, and source of coverage. © 2016 The Authors. Health Economics published by John Wiley & Sons Ltd.


STUDY DATA AND METHODS
We used data from the ACS, conducted by the Census Bureau, to examine insurance changes between 2013 and 2014. The ACS is an excellent data source because of its large sample size, mandatory respondent participation (unlike the Current Population Survey (CPS) or the Behavioral Risk Factor Surveillance System (BRFSS)), breadth of questions on sources of insurance, focus on contemporaneous coverage, and uniformity of questions over time. 1 Unlike the BRFSS, the ACS has several distinct categories of health insurance sources. The ACS focuses explicitly on current coverage, leading to less confusion than the CPS questions, which ask about coverage in the previous calendar year. The CPS question is sometimes misinterpreted by respondents as asking about current coverage. 2 Importantly, the CPS insurance questions were redesigned in 2014, increasing the difficultly in making over-time comparisons. 3 One drawback with the public version of the ACS is that the respondent's interview date within the year is unknown. However, virtually all changes from the ACA occurred on January 1, limiting the need for precise interview timing. 4 We classify those who were elderly (or disabled) and report being on both Medicare and another plan as Medicare beneficiaries. We distribute the small remaining fraction with multiple sources of coverage evenly across reported sources of coverage, such that the totals add up to 100%. None of these assumptions affect our calculation of insurance gains.
One confounding factor is that the national economy was improving. The unemployment rate fell from 8.0 to 5.6% between January 2013 and December 2014, and it also fell in every state. 5 An improving labor market would likely lead to gains in employer-provided coverage.

POTENTIAL FOR GAINS IN COVERAGE BECAUSE OF THE ACA
The ACS measures household income relative to the FPL and the respondent's age. As is illustrated in Table I, the federal ACA provisions, along with state choices, imply different potential sources of coverage gains for individuals depending on their age and income. For the elderly, we expect very little change, because virtually all qualify for Medicare. Poor children were virtually all previously eligible for Medicaid. 6 However, the individual mandate may have compelled parents to enroll their children who were previously "conditionally covered". 7 For children above the FPL, the mandate and other insurance market changes would be expected to raise coverage.
For non-elderly adults, coverage gains and sources are expected to vary depending on income. In nonexpansion states, poor adults were not entitled to a federal subsidy, but could purchase community-rated individual Marketplace coverage. 8 In Medicaid expansion states, such individuals would be eligible for Medicaid. Thus, the majority of the expected increase in coverage for poor adults would likely come from Medicaid in expansion states and private coverage in non-expansion states. Given differences in premiums, we expect the increase in coverage in expansion states to be larger for poor adults.
Adults with incomes falling between 100 and 137% of the FPL were eligible for Medicaid in expansion states and were eligible for large federal subsidies for Marketplace coverage in non-expansion states. 9 Thus, one would expect sizable gains for both groups, but from different sources. For those with incomes falling between 138 and 399% of the FPL, Marketplace purchases were subsidized on a sliding scale, so gains should come from private coverage. Finally, for those with higher incomes, the individual mandate compels coverage, but purchases were not subsidized. An individual for whom insurance was deemed affordable could elect not to purchase coverage and instead pay a penalty, which was relatively low in 2014. 10 9 For an adult at the FPL, the expected contribution towards coverage for the second lowest silver plan would be 2 percent of income. The voucher amount could then be applied to any plan. 10 https://www.healthcare.gov/fees/fee-for-not-being-covered/  The subsequent tables characterize actual changes in coverage levels and sources between 2013 and 2014 and compare those changes to expectations described earlier. Table II lists the percentage uninsured nationally and by state, as well as the percentage covered by each source for 2013 and 2014. Nationally, the uninsured rate fell from 14.8 to 12.0%, a 2.8 percentage point reduction that was divided equally between gains in Medicaid and individual coverage. The largest gains occurred in two states that opted to expand Medicaid, Kentucky (5.7 percentage points) and Nevada (5.2 percentage points). These increases were largely driven by gains in Medicaid coverage. Kentucky experienced a 4.3 percentage point increase in Medicaid coverage, and Nevada experienced a 3.7 percentage point increase. The smallest coverage gains occurred in Wyoming and Massachusetts. Factors ranging from the choice not to expand Medicaid, strong local economies, and prior health insurance reforms likely explain the modest gains for states such as these. 11,12 Table III aggregates coverage into three categories (uninsured, public, and private coverage) and stratifies by age and income. Among age groups, the largest coverage gains occurred for those aged 19 to 25 (5.3 percentage points). These young adults had a 1.9 percentage point increase in public coverage and a 3.4 percentage point increase in private coverage. The large gains among young adults are interesting because they already received a sizeable insurance expansion in 2010 with the mandate for private insurers to allow dependents to remain on parents' coverage until age 26. 13 The largest gain by income category occurred for those with incomes between 100 and 137 of the FPL (5.5 percentage points). The gain in public (private) coverage for this group was 3.5 (1.9) percentage points. There was no change in coverage for the elderly.  Table I suggests that we expect the biggest potential gains in coverage to come from near-poor non-elderly adults, although the source is expected to vary based on state Medicaid expansion status. Table IV examines this by stratifying by expansion status. There were increases in coverage of 8.8, 10.5, and 7.5 percentage points for adults with income below the FPL, between 100 and 137% of the FPL, and between 138 and 249% of the FPL, respectively, in expansion states. In non-expansion states, the increases in coverage within these same groups are 4.7, 7.0, and 4.7 percentage points. Thus, the coverage gains were larger in expansion states within each of these income categories. 14 As predicted, the majority of the increase in coverage for these adults in the expansion states came from public coverage, while the majority of the gain in coverage for these adults in nonexpansion states came from private coverage. This is true even for adults in the 138 to 249% of the FPL range, where we might have expected gains in private coverage to outweigh gains in public coverage in all states.

RESULTS
For higher income adults, we still see gains in coverage in both expansion and non-expansion states despite decreasing federal subsidies that phase-out by 400% of the FPL. These adults may be responding to the mandate or benefiting from the other insurance market reforms. These increases were generally higher in expansion states. We observe gains in coverage in both expansion and non-expansion states for children, especially poor children. Among poor children who were already Medicaid-eligible, coverage increased by 1.7 (1.3) percentage points in expansion (non-expansion) states. These increases are driven by increased formal participation in Medicaid, as opposed to 'conditional coverage', in which parents might wait until their Medicaid-eligible child needs care to formally enroll them. This change could potentially be attributed to the mandate. We also see gains in coverage for children in families whose income (100-249% of the FPL) would have likely already qualified them for CHIP coverage.

DISCUSSION
A recent analysis of National Health Interview Survey focuses only on young adults and finds significant gains in coverage. 15 Our findings are consistent with this, although our purpose is to consider both broad national trends for all ages, as well as differences by state, age, income, and source of coverage. Overall, our results are generally consistent with our expectations regarding where the pieces of the ACA implemented in 2014 would have the largest impact.
The 2.8 percentage point national gain in coverage we observe was divided equally between gains in Medicaid and private coverage. We observe larger gains in Medicaid coverage for low-income non-elderly adults in states made newly eligible for Medicaid, while we see larger gains in private coverage for lowincome non-elderly adults eligible for federal subsidies in non-expansion states. We see gains in public coverage even among children whose family income already qualified them for public coverage.
Approximately 12% of the US population remained uninsured in 2014. One might expect additional gains in coverage as the economy continues to improve, as more states expand Medicaid, and as the size of the uninsured penalty increases. However, administration changes in several states, including most notably Kentucky, may cause states to revisit their expansion decisions or future plans. Thus, complete and timely data on changes in patterns of coverage will continue to be of interest to policymakers.