Taking Stock: Health Insurance Coverage under the ACA as of March 2016
Michael Karpman, Sharon K. Long, and Stephen Zuckerman
May 25, 2016
The Urban Institute has used its Health Reform Monitoring Survey (HRMS) to examine trends in health insurance coverage since the first quarter of 2013. The HRMS is one of several surveys, including those conducted by federal agencies and other private research organizations, that have found a decline in the share of nonelderly adults without health insurance since the implementation of the key coverage provisions of the Affordable Care Act, or ACA (Collins et al. 2015; Karpman and Long 2015; Smith and Medalia 2015; Uberoi, Finegold, and Gee 2016; Zammitti and Cohen 2016).1 Several recent studies have attributed most of the reduction in uninsurance to the ACA (Blumberg, Garrett, and Holahan 2016; Courtemanche et al. 2016; Frean, Gruber, and Sommers 2016).
In this brief, we provide updated estimates of changes in coverage between September 2013, just before the first Marketplace open enrollment period, and March 2016, just after the end of the third open enrollment period. We focus on changes in the share of adults without coverage at the time of the survey and changes in continuity of coverage, as measured by the share of adults who reported having health insurance for all 12 months before the survey. We find a sustained decline in the uninsurance rate and an increasing share of adults with full-year coverage, with states that implemented the ACA’s Medicaid expansion seeing the greatest gains in coverage and continuity of coverage, particularly among low-income adults.
What We Did
We examine changes in the uninsurance rate and in the rate of full-year insurance coverage2 for nonelderly adults (ages 18 to 64) overall and by state Medicaid expansion status.3 We also report changes by family income as a percentage of the federal poverty level (FPL), with emphasis on those targeted by Medicaid expansion (incomes at or below 138 percent of FPL) and those potentially eligible for subsidized Marketplace coverage (incomes between 139 and 399 percent of FPL).4 We focus on changes since September 2013 that are significantly different from zero at the 5 percent level. We provide a 95 percent confidence interval (CI) for key estimates. The size of our analytic sample is 7,911 for the September 2013 round and 7,520 for the March 2016 round.
Estimates from the HRMS are weighted to be nationally representative. We use weights and regression adjustment5 to control for differences in the demographic and socioeconomic characteristics of respondents across the different rounds of the survey. This allows us to remove changes in insurance coverage caused by changes in the types of respondents over time rather than by changes in the health insurance landscape. The basic patterns shown for the regression-adjusted measures are similar to those based solely on simple weighted estimates. In presenting the regression-adjusted estimates, we use the predicted rate of coverage in each quarter for the same nationally representative population. For this analysis, we base the nationally representative sample on survey respondents for the four most recent rounds of the survey for quarters 1 and 3 to match the current timing of the HRMS.6
This analysis has several limitations. Though HRMS estimates capture changes in insurance coverage from the first ACA open enrollment period, estimates will not capture the full effects of the ACA because they do not reflect the effects of some important ACA provisions implemented before 2013 (e.g., early state Medicaid expansions and the ability to keep dependents on health plans until age 26). In addition, these change estimates reflect more than the effects of the ACA because they do not control for long-term trends in health insurance coverage that predate the ACA, changes in the business cycle, or other factors.7 Further, the difference in coverage gains between Medicaid expansion states and nonexpansion states should not be entirely attributed to Medicaid expansion. Other factors, including differences in other ACA policy choices, likely affected enrollment. For example, many nonexpansion states did not set up their own Marketplaces and therefore did not receive the same funding for outreach and enrollment assistance.
What We Found
The uninsurance rate for nonelderly adults fell from 17.6 percent in September 2013 to 9.9 percent in March 2016, a 43.8 percent decline representing 15.5 million fewer uninsured adults.
Between September 2013 and March 2016, the uninsurance rate for nonelderly adults fell 7.7 percentage points (95 percent CI [6.5, 9.0]) from 17.6 percent to 9.9 percent (figure 1). Based on an estimate of the 2016 population, this represents 15.5 million fewer uninsured adults (95 percent CI [13.0 million, 18.0 million]) in 2016, a decrease from 35.4 million (95 percent CI [29.5 million, 41.4 million]) based on the 2013 uninsurance rate to 19.9 million (95 percent CI [14.6 million, 25.2 million]) based on the 2016 uninsurance rate.8
As expected, much of the decline in uninsurance occurred during the first and second open enrollment periods. Following the third open enrollment period, we find a small decline between September 2015 and March 2016 of 0.5 percent percentage points (95 percent CI [-0.3, 1.4]), which, although not statistically significant, is similar to the Gallup estimate of a 1.3 percentage-point decline between quarter 3 2015 and quarter 1 2016.9 HRMS and Gallup estimates suggest a recent slowing of the decline in uninsurance relative to the period immediately following implementation of the ACA’s key coverage provisions.
As uninsurance has dropped over time, continuity of insurance coverage has increased.
By expanding eligibility for no-cost or subsidized coverage to more adults and imposing a tax penalty for going without coverage, the ACA is expected to increase the share of adults who gain and keep coverage over time, providing better health care access and affordability (Haley and Zuckerman 2003).10 We find an increase, consistent with expectations for the ACA, in the share of adults with full-year coverage between September 2013 and March 2015, about one year after the end of the first open enrollment period, and a continued increase in the rate of full-year coverage through March 2016 (figure 2). By March 2016, 82.0 percent of adults reported that they had insurance for all 12 months before the survey, up from 75.0 percent in September 2013.
The greatest reductions in uninsurance and the greatest gains in full-year coverage were reported by adults, particularly low-income adults, in Medicaid expansion states.
As shown in figure 1, uninsurance among adults in states that had expanded Medicaid by January 2016 fell 8.5 percentage points (95 percent CI [6.9, 10.1]), from 15.8 percent in September 2013 to 7.3 percent in March 2016, a decline of 53.8 percent. Among adults in nonexpansion states, the uninsurance rate fell 6.5 percentage points (95 percent CI [5.1, 7.9]), from 20.6 percent to 14.1 percent, a 31.5 percent decline.11 As a result, the uninsurance rate for adults in nonexpansion states was nearly double that of adults in expansion states in March 2016.
At the same time, the increase in continuous coverage was significantly greater in expansion states than in nonexpansion states. The share of adults with continuous coverage rose from 76.8 percent in September 2013 to 85.4 percent in March 2016 in expansion states, a gain of 8.6 percentage points (95 percent CI [7.4, 9.9]). Nonexpansion states saw a 4.6 percentage-point gain (95 percent CI [3.0, 6.1]), from 72.1 percent to 76.7 percent (figure 2).
The larger increase in continuity of coverage in expansion states likely reflects the stronger safety net provided by the Medicaid expansion, which is more likely to reach the existing pool of low-income adults without coverage and “catch” more low-income adults who lose or can no longer afford private coverage. Supporting that hypothesis, we see particularly strong gains in coverage and continuity of coverage for low-income adults in expansion states. Between September 2013 and March 2016, there was a 19.4 percentage-point decrease (95 percent CI [14.8, 23.9]) in the share of adults with incomes at or below 138 percent of FPL in expansion states who were uninsured at the time of the survey, from 35.3 percent to 16.0 percent (figure 3). Nonexpansion states saw a decrease in the share of low-income adults without coverage of just 11.2 percentage points (95 percent CI [7.2, 15.1]) over the same period.12 At the same time, there was a 20.9 percentage point increase (95 percent CI [16.7, 25.0]) in the share of low-income adults with full-year coverage in the expansion states, compared with a 10.1 percentage point increase (95 percent CI [4.1, 16.1]) in nonexpansion states (figure 4).
Among adults targeted by the ACA’s Marketplace tax credits (i.e., those with incomes between 139 and 399 percent of FPL), the percentage-point decrease in uninsurance was about the same in expansion and nonexpansion states. Full-year coverage increased among these moderate-income adults in expansion states; in nonexpansion states, the change was not statistically significant.
What It Means
These results indicate sustained progress under the ACA in reducing the share of uninsured nonelderly adults and increasing the share of adults with continuous coverage. An estimated 15.5 million fewer adults were uninsured in March 2016 following the implementation of the ACA’s key coverage provisions, as the uninsured rate fell 43.8 percent from September 2013. Over the same period, the share of adults with full-year coverage rose from 75.0 percent to 82.0 percent, with large gains in continuous coverage and reductions in uninsurance among low-income adults in Medicaid expansion states. Although the rate of decline in the uninsurance rate has slowed in the past year, only 1 in 10 adults remain uninsured as of March 2016.
With the much higher levels of uninsurance among low-income adults in states that did not expand Medicaid, it is likely that many of the remaining uninsured are in the coverage gap (Blumberg et al. 2016; Garfield and Damico 2016). In nonexpansion states, adults are ineligible for Medicaid or financial assistance to obtain coverage if their income is above the state’s Medicaid eligibility cutoff and below 100 percent of FPL; they are not poor enough to qualify for Medicaid but too poor to receive Marketplace tax credits. Previous studies indicate that low- and moderate-income adults may move in and out of this coverage gap as their incomes fluctuate over time (Buettgens, Nichols, and Dorn 2012; Sommers and Rosenbaum 2011; Sommers et al. 2014).
In contrast, the expansion of Medicaid eligibility to those with incomes at or below 138 percent of FPL seems to have created a wider coverage safety net for adults who lose a job or face a decline in income, thereby preventing an interruption in coverage. Nearly three-quarters of low-income adults in expansion states reported full-year coverage in March 2016 compared with just over half of low-income adults in nonexpansion states. Low-income insured adults in expansion states were also more likely to have retained coverage for a full year than those in nonexpansion states (data not shown). Other ACA rules may also mitigate churn across eligibility levels by providing more stable measures of income. These include the state options to count predictable changes in income when calculating current monthly income to determine Medicaid eligibility and to use projected annual income based on the prior year when determining continued eligibility among existing beneficiaries (Artiga, Musumeci, and Rudowitz 2012; Brooks 2015). More directly, New York and Montana have adopted 12-month continuous eligibility for adults eligible for Medicaid, ensuring continuous coverage over the year.13
The slowdown in coverage gains between the second and third open enrollment periods observed in the HRMS and Gallup surveys suggests that many of the remaining adults who could benefit from the ACA are either not aware of the benefits, live in states that have not made the Medicaid expansion available to them, or feel coverage is still unaffordable. Additional targeted outreach to inform people of ACA benefits, combined with more enrollment assistance, may lead to some further gains in coverage. The higher individual mandate penalties imposed in 2016 may also encourage more enrollment in coming years. Still, more substantial gains in coverage will require extending the Medicaid expansion to more states and increasing the availability and/or level of financial assistance, both of which would require state or federal action.
Artiga, Samantha, MaryBeth Musumeci, and Robin Rudowitz. 2012. Medicaid Eligibility, Enrollment Simplification, and Coordination under the Affordable Care Act: A Summary of CMS’s March 23, 2012 Final Rule. Washington, DC: Kaiser Commission on Medicaid and the Uninsured.
ASPE (Office of the Assistant Secretary for Planning and Evaluation). 2015. Health Insurance Marketplace 2015 Open Enrollment Period: March Enrollment Report. Washington, DC: US Department of Health and Human Services.
Blumberg, Linda J., Bowen Garrett, and John Holahan. 2016. “Estimating the Counterfactual: How Many Uninsured Adults Would There Be Today without the ACA?” Inquiry 53. doi: 10.1177/0046958016634991.
Blumberg, Linda J., Michael Karpman, Matthew Buettgens, and Patricia Solleveld. 2016. Who Are the Remaining Uninsured, and What Do Their Characteristics Tell Us About How to Reach Them? Washington, DC: Urban Institute.
Brooks, Tricia. 2015. Getting MAGI Right: A Primer on Differences that Apply to Medicaid and CHIP. Washington, DC: Georgetown University Health Policy Institute.
Buettgens, Matthew, Austin Nichols, and Stan Dorn. 2012. Churning under the ACA and State Policy Options for Mitigation. Washington, DC: Urban Institute.
CMS (Centers for Medicare and Medicaid Services). 2016. Medicaid & CHIP: December 2015 Monthly Applications, Eligibility Determinations and Enrollment Report. Baltimore, MD: US Department of Health and Human Services.
Collins, Sara R., Petra W. Rasmussen, Michelle M. Doty, and Sophie Beutel. 2015. Americans’ Experiences with Marketplace and Medicaid Coverage: Findings from the Commonwealth Fund Affordable Care Act Tracking Survey, March–May 2015. New York: Commonwealth Fund.
Courtemanche, Charles, James Marton, Benjamin Ukert, Aaron Yelowitz, and Daniela Zapata. 2016. “Impacts of the Affordable Care Act on Health Insurance Coverage in Medicaid Expansion and Non-Expansion States.” Working Paper 22182. Cambridge, MA: National Bureau of Economic Research.
Frean, Molly, Jonathan Gruber, and Benjamin D. Sommers. 2016. “Premium Subsidies, The Mandate, and Medicaid Expansion: Coverage Effects of the Affordable Care Act.” Working Paper 22213. Cambridge, MA: National Bureau of Economic Research.
Garfield, Rachel, and Anthony Damico. 2016. “The Coverage Gap: Uninsured Poor Adults in States that Do Not Expand Medicaid — An Update.” Menlo Park, CA: Kaiser Family Foundation.
Haley, Jennifer, and Stephen Zuckerman. 2003. Is Lack of Coverage a Short- or Long-Term Condition? Washington, DC: Kaiser Commission on Medicaid and the Uninsured.
Karpman, Michael, and Sharon K. Long. 2015. QuickTake: Taking Stock: Gains in Health Insurance Coverage under the ACA Continue as of September 2015, but Many Remain Uninsured. Washington, DC: Urban Institute.
Smith, Jessica C., and Carla Medalia. 2015. Health Insurance Coverage in the United States: 2014. Washington, DC: US Census Bureau.
Sommers, Benjamin D., and Sara Rosenbaum. 2011. “Issues in Health Reform: How Changes in Eligibility May Move Millions Back and Forth Between Medicaid and Insurance Exchanges.” Health Affairs 30 (2): 228–36.
Sommers, Benjamin D., John A. Graves, Katherine Swartz, and Sara Rosenbaum. 2014. “Medicaid and Marketplace Eligibility Changes Will Occur Often in All States; Policy Options Can Ease Impact.” Health Affairs 33 (4): 700–707.
SHADAC (State Health Access Data Assistance Center). 2013. Comparing Federal Government Surveys that Count the Uninsured. Minneapolis, MN: State Health Access Data Assistance Center.
Uberoi, Namrata, Kenneth Finegold, and Emily Gee. 2016. Health Insurance Coverage and the Affordable Care Act, 2010–2016. Washington, DC: US Department of Health and Human Services.
Zammitti, Emily P., and Cohen, Robin A. 2016. Health Insurance Coverage: Early Release of Quarterly Estimates from the National Health Interview Survey, January 2010-December 2015. Hyattsville, MD: Centers for Disease Control and Prevention.
About the Series
This brief is part of a series drawing on the HRMS, a survey of the nonelderly population that explores the value of cutting-edge Internet-based survey methods to monitor the ACA before data from federal government surveys are available. Funding for the core HRMS is provided by the Robert Wood Johnson Foundation and the Urban Institute.
For more information on the HRMS and for other briefs in this series, visit www.urban.org/hrms.
About the Authors
Michael Karpman is a research associate, Sharon K. Long is a senior fellow, and Stephen Zuckerman is a senior fellow and codirector with the Urban Institute's Health Policy Center. The authors gratefully acknowledge the suggestions and assistance of Linda J. Blumberg, Katherine Hempstead, and John Holahan.
1 Stephanie Marken, “U.S. Uninsured Rate at 11.0%, Lowest in Eight-Year Trend,” Gallup, April 7, 2016.
2 Information is missing on coverage status for the 12 months before the survey for less than 1 percent of respondents in each round of the HRMS. When reporting the share of adults with continuous coverage, we assign these people as having less than full-year coverage.
3 We focus on changes in insurance coverage because estimates of the level of coverage often vary across survey programs because of differences in the surveys that are unrelated to the ACA (SHADAC 2013). For this analysis, we focus on state decisions to expand Medicaid by January 1, 2016. The states that expanded Medicaid by this date are AK, AZ, AR, CA, CO, CT, DE, DC, HI, IL, IN, IA, KY, MD, MA, MI, MN, MT, NH, NV, NJ, NM, NY, ND, OH, OR, PA, RI, VT, WA, and WV. Several of those states, including CA, CT, DC, and MN, expanded Medicaid under the ACA before 2013. Among nonexpansion states, WI has used state funding to expand eligibility to nonelderly adults with incomes up to 100 percent of FPL, and LA is planning to expand eligibility for Medicaid by mid-2016. Because our definition of states expanding Medicaid changes as more states adopt the expansion, estimates based on state Medicaid expansion status are not necessarily comparable over time.
4 The income measure used in the HRMS is not the same as the measure of modified adjusted gross income used to determine eligibility for Medicaid and Marketplace tax credits. Therefore, estimated levels of family income as a percentage of FPL only provide an approximation of whether respondents are in the income ranges targeted by the Medicaid expansion and Marketplace tax credits.
5 Specifically, we control for the variables used in the poststratification weighting of the KnowledgePanel (the Internet-based survey panel that underlies the HRMS) and the poststratification weighting of the HRMS. These variables are sex, age, race and ethnicity, language, education, marital status, presence of children in the household, household income, family income as a percentage of FPL, homeownership status, Internet access, urban or rural status, and census region. In this analysis, we also control for citizenship status and participation in the previous quarter’s survey (i.e., whether the responded completed the survey in the previous quarter, was sampled in the previous quarter but did not complete the survey, or was not sampled in the previous quarter). Beginning in September 2015, we shifted to a more parsimonious regression model to better support subgroup analyses. Specifically, we collapsed detailed categories for some covariates and dropped some interaction terms.
6 Given the switch from a quarterly fielding schedule to a semiannual schedule in March 2015, we now use data from the most recent four rounds of the HRMS that are fielded in March and September to predict uninsurance rates for the same nationally representative population rather than data from the most recent 12-month period. These changes, first implemented with the analysis of the September 2015 HRMS data (Karpman and Long 2015), have little effect on the regression-adjusted estimates.
7 See Blumberg, Garrett, and Holahan (2016) for an approach that uses HRMS and Current Population Survey data to isolate changes in coverage that occurred because of the ACA from changes that can be explained by long-term trends and the business cycle.
8 We use projections for the size of the 2016 population from the US Census Bureau. These files give population projections by race, ethnicity, and sex for all ages from 2014 to 2060 based on estimated birth rates, death rates, and net migration rates. Using table 1, which has a 2016 projected population of 323,995,528, we summed the 2016 population projections for all 18- to 64-year-olds to arrive at 200,873,038 nonelderly adults. See “2014 National Population Projections: Downloadable Files,” US Census Bureau, last updated April 21, 2016.
9 For the Gallup estimate for quarter 1 2016, see Stephanie Marken, “U.S. Uninsured Rate at 11.0%, Lowest in Eight-Year Trend,” Gallup, April 7, 2016. For quarter 3 2015 estimates, see Stephanie Marken, “U.S. Uninsured Rate at 11.6% in Third Quarter,” Gallup, October 8, 2015.
10 The Incidental Economist, “The Consequences of Health Insurance Coverage Gaps,” AcademyHealth, January 6, 2014.
11 The decline in uninsurance in expansion states was significantly different from the decline in nonexpansion states at the 0.10 level.
12 Although many adults with incomes below 138 percent of FPL in nonexpansion states do not qualify for financial assistance to obtain coverage, those with incomes between 100 and 138 percent of FPL are eligible to receive Marketplace tax credits, and some Medicaid-eligible parents with very low incomes may have recently taken up coverage. In nonexpansion states that use the federal Marketplace platform, half of the 6 million individuals enrolled as of February 2015 with available data on income reported incomes below 150 percent of FPL (ASPE 2015). There were also more than 2 million additional individuals in nonexpansion states enrolled in Medicaid or CHIP in December 2015 compared to the period July–September 2013 (CMS 2016). While some of these Marketplace and Medicaid enrollees may have had other coverage prior to enrollment, it is likely that many were previously uninsured.
13 “State Adoption of 12-Month Continuous Eligibility for Children’s Medicaid and CHIP,” Kaiser Family Foundation, accessed May 16, 2016.