The report was written by
* * *
* Federal surveys relied on by researchers and policymakers for estimates of health insurance coverage have been disrupted by the COVID-19 pandemic, potentially influencing the accuracy of their estimates.
* Recent survey data suggest a stable uninsured rate in 2020 despite the pandemic and related recession, with estimates ranging from 8.6 percent to 9.7 percent of the population (28.0 million to 31.6 million people).
* Examination of other available data sources provides insights about changes in coverage by source and the overall stability of insurance rates in 2020. Administrative data over the same period suggest that increases in Medicaid and Marketplace enrollment roughly offset decreases in employer coverage, potentially explaining the survey findings.
* More recent administrative data in 2021 show ongoing gains in Medicaid (1.7 million between January and
* Uninsured rates continue to be higher in certain populations, including Latinos (18.3 percent) and Blacks (10.4 percent), people with incomes below the poverty level (17.2 percent), and residents of states that have not expanded Medicaid (17.6 percent).
* These results can inform Open Enrollment efforts and strategies being taken to expand coverage further in 2022.
* * *
The COVID-19 pandemic and its economic fallout raised concerns that the number of uninsured individuals would increase in 2020, with higher unemployment in 2020 potentially causing significant reductions in employer coverage. Federal surveys estimated that approximately 30 million
Understanding the trends in health insurance coverage during the months of the pandemic and the impacts of recent policy changes can guide future decision-making.
Efforts to understand the health insurance dynamics of the pandemic are complicated by the fact that the pandemic also created challenges in conducting government-administered surveys that provide the most robust measurement of insurance coverage./2
As a result, survey measurement of insurance may have more uncertainty than normal, though the magnitude of these impacts are not yet known./3
The purpose of this Issue Brief is to highlight challenges using federal survey data for estimates on health coverage and uninsurance in 2020 and early 2021; examine alternative data sources for insights into coverage changes; and summarize what is known to date about health coverage and uninsurance during this period.
FEDERAL SURVEY CHALLENGES DURING COVID-19 PANDEMIC
The mode of data collection and timing of each federal survey influences how its administration was affected by the COVID-19 pandemic. Table 1 below summarizes the nationwide surveys used to estimate health insurance coverage, their typical mode of collection, and impact of the pandemic on survey administration.
* * *
[See link at end of text for Table 1: Summary of Federal Health Insurance Surveys During COVID-19 Pandemic]
* * *
By design, the ACS is a continuously fielded survey and collects data for a new sample each month. Due to the pandemic, from mid-March through
It noted that the "2020 ACS data collection had the lowest response rate ever for the survey at 71 percent, down from 86 percent in 2019 and 92 percent in 2018." In addition to the reduced response rate, there was evidence that those who did respond differed systematically from those who did not, in terms of social, economic and housing characteristics, raising concerns that 2020 estimates suffer from nonresponse bias. More specifically, nonresponse was more common among people with lower incomes, lower educational attainment, and who were less likely to own their home.
Along with the experimental weights and 2020 ACS 1-year Public Use Microdata Sample (PUMS) file, the
Current Population Survey Annual Social and Economic Supplement (CPS ASEC)
The CPS ASEC surveys roughly 100,000 addresses and asks about coverage for each month in the prior year. Roughly 75 percent of these interviews occur in March./8
While it is a much smaller survey than the ACS, the CPS ASEC captures socioeconomic data in more detail, particularly relating to income and receipt of government benefits.
The 2020 CPS ASEC asked respondents to report their health insurance coverage for 2019. On
In describing the 2020 CPS ASEC's estimates of 2019 coverage, a
For the 2021 CPS ASEC, fielded February -
Again, there are concerns that those who did respond differed meaningfully from those who did not. In an analysis of the W-2 earnings of households responding to the survey in 2020 and 2021, the
An additional change made in the NHIS for 2020 was the addition of a follow-back component in the second half of 2020, in which a portion of 2019 NHIS respondents were re-interviewed in 2020. In an analysis of survey respondents in Q2 compared to Q1, NCHS staff found that higher socio-economic status households were over-represented in Q2. The Q2 2020 early release estimates used modified weighting procedures to attempt to adjust for measurable differences in respondents as a result of the survey mode change. In the most recent survey release, NCHS noted that estimates for 2020 may still be affected by these changes./13
In 2019 the response rate for the private sector sample was 59.2 percent; in 2020 the equivalent response rate was 56.1 percent. While this is a slight decrease, response rates have been steadily declining in the survey since 2012 and the 2020 figure does not represent a dramatic departure from that trend./14
Because the MEPSIC surveys employers rather than households, it may have been less affected by the pandemic than surveys that typically interview households in-person. However, because unemployment changed significantly during the year, estimates of employment and insurance coverage may be correlated by timing of response. AHRQ is investigating whether these factors influence estimates from the 2020 MEPS-IC data./15
A challenge with HPS data is that it has a high non-response rate overall, as well as for certain individual questions. The overall response rate is roughly 7 percent./16
Even among respondents, roughly 20 percent of respondents fail to complete all 8 health insurance questions, which forces researchers to make assumptions about the nature of non-respondents. Despite these limitations, the HPS is the most recent federal survey data source available and is frequently updated.
SURVEY ESTIMATES OF UNINSURED POPULATION IN 2020 AND EARLY 2021
Figure 1 presents the trends in non-elderly adult* uninsured rates 2018-2020 estimated from federal surveys. CPS ASEC estimates of insurance are "ever uninsured" during the year and, as a result, the survey's estimates of uninsurance are lower than the point-in-time estimates for the NHIS and ACS.
* * *
[See link at end of text for Figure 1: Federal Survey Estimates of Uninsured Rates Among Non-Elderly Adults, 2018-2020]
* * *
The most recently published early release NHIS data extend through Q1 2021. They, along with the full year 2020 data, show that the non-elderly adult uninsurance rate (13.9 percent) was lower than, but not significantly different from, 2019 (14.7 percent). Overall, 31.6 million persons of all ages were uninsured at the time of interview in 2020; this estimate is lower than, but not significantly different from, 2019 where 33.2 million persons of all ages were uninsured. While there is potential for nonresponse bias in these estimates, they do not on the whole show evidence for an increased uninsured rate in 2020 and early 2021.
Current Population Survey Annual Social and Economic Supplement (CPS ASEC) In
The survey did find that the composition of insurance coverage changed somewhat, with increasing numbers of individuals with public coverage and fewer with private coverage, compared to 2018. Notably, the survey found no change in the number with Medicaid even though administrative data (discussed below) show increased Medicaid enrollment during this period. One possible explanation for this discrepancy is that the CPS ASEC uses an "ever-on" measure for health insurance; this means that those already on Medicaid in January or
Nonresponse bias not fully accounted for by survey weights may have led to a further undercount of Medicaid enrollment changes during this period.
The CPS ASEC estimates also indicate substantial disparities in coverage. Uninsured rates in 2020 were significantly higher among Latinos (18.3 percent) and Blacks (10.4 percent) compared to Asians (5.9 percent) and Whites (5.4 percent). People with incomes below the federal poverty level (FPL) experienced an uninsured rate of 17.2 percent, compared to 3.4 percent among those with incomes above 400 percent of FPL. Residents of states that have not expanded Medicaid were nearly twice as likely to be uninsured compared to those living in expansion states (17.6 percent vs. 8.9 percent).
A study using HPS data found that rates of employer coverage declined throughout 2020 (particularly in the first months of the pandemic), while enrollment in other types of health insurance coverage increased./19
These findings are generally consistent with results from the CPS and NHIS, which suggest that public programs played an important role in offsetting declines in employer coverage during 2020.
Non-Government Survey Estimates
The HRMS completion rate is roughly 5 percent./21
Panel members agree to participate in regular surveys, including the HRMS, and typically stay in the panel for two years. Recent field dates were
ADMINISTRATIVE DATA AND OTHER SOURCES
Unlike survey data, administrative data on coverage enrollment was generally not subject to increased data quality concerns during the pandemic. However, administrative data can only capture information on who has specific types of coverage and cannot provide insight about those who lack health insurance coverage, which typically must come from surveys. Administrative data on Medicare, Medicaid, and Marketplace enrollment can provide enrollment trends in these programs during the pandemic. There is not, however, a comprehensive administrative data source on employer coverage, though there are surveys of employers that provide information in this area. Data from health plans' financial filings can also provide useful information. This section examines these data sources and what is known about these coverage types in 2020 and early 2021.
(Appendix Table 1). Three states -
Moreover, starting in
Based on previous recessions' effect on enrollment in the Medicaid program, Medicaid enrollment shifts typically lag behind other social supports (e.g., unemployment insurance,
Early in the COVID-19 pandemic, multiple research groups projected changes in health coverage and Medicaid enrollment based on unemployment rates and the economic recession. Estimated increases in Medicaid enrollment ranged from 530 million individuals./28,29,30,31,32
Most of these initial estimates were higher than reported increases in Medicaid enrollment to date. Once administrative data from the Spring of 2020 became available, CMS and other researchers found the increase in Medicaid enrollment was not driven by the increase in new applications (for example, from individuals who lost their job/coverage and became Medicaid-eligible); in fact, the growth in enrollment outpaced the increase in new applications./33,34
Instead, the growth in Medicaid enrollment was largely driven by a requirement for states to suspend eligibility redeterminations in order to receive a 6.2 percentage point increase in their FMAP per the FFCRA provision. To further support states in their response to COVID-19, CMS developed numerous strategies and flexibilities to support Medicaid and CHIP operations during this time, often resulting in expedited enrollment and retention (e.g., presumptive eligibility, continuous eligibility, waiving premiums and cost sharing, regulatory authority to apply exceptions to the timeliness standards for application and renewal processing). Additional research found reported growth in Medicaid enrollment was largest in states that experienced smaller increases in unemployment rate./35
In addition to the research community, the health insurance industry also expected commercial enrollment to decrease and Medicaid enrollment to grow in response to the high unemployment rate in Spring 2020. Although Medicaid enrollment increased during 2020, the increase was smaller than the industry expected. Possible explanations for the lower increase include that those who lost their jobs at the beginning of the pandemic were more likely to be uninsured already, coverage provisions in the ACA played a critical role helping people maintain and gain coverage, economic support from FFCRA and the CARES Act helped stimulate economic recovery, and many employers opted to furlough rather than terminate their employees to maintain their benefits (e.g., health insurance)./36
The nongroup insurance market consists of on- and off-Marketplace individual insurance coverage, as well as grandfathered and transitional policy plans that are not subject to most ACA reforms. Marketplace insurance is purchased through the federal HealthCareHealthCare.gov platform and State-based Marketplaces (SBMs). Off-Marketplace insurance is by purchased directly from insurers. Federal advanced premium tax credits (APTCs) are available for Marketplace coverage, depending on a person's income. A large majority of Marketplace enrollees received premium tax credits (87 percent of enrollees in
Effectuated enrollment was 10.6 million in
* * *
[See link at end of text for Table 2: Marketplace Enrollment
The American Rescue Plan Act of 2021 (ARP) enhances and expands eligibility for APTCs to Marketplace consumers for 2021 and 2022. Under the ARP, an estimated 79 percent of HealthCare.gov enrollees became eligible for health plans with zero premiums, and 87 percent became eligible for low premium plans (
The ARP also offers enhanced 2021 Marketplace subsidies to consumers who are in households receiving unemployment compensation for any week in 2021./40
The federal Marketplace on the HealthCare.gov platform for 36 states offered a COVID-19 Special Enrollment Period (SEP) from
Without the COVID-19 SEP, consumers could only enroll in the Marketplace if they had a qualifying life event outside the open enrollment period (OEP). These ARP-enhanced APTCs led to increased Marketplace enrollment during the SEP. There were 2.8 million total SEP plan selections for HealthCare.gov and SBM states
Overall, a total of 12.2 million people had effectuated Marketplace coverage in
Individual nongroup coverage is also available off-Marketplace for minimum essential coverage. Estimates of the number of persons enrolled in off-Marketplace individual coverage (which includes non-ACA compliant coverage such as grandfathered and grandmothered plans) range between 3.7 and 5 million in 2019, but we do not have estimates for 2020./43,44,45
Employer Coverage and Other Group Health Plans
Multiple data sources exist related to employer coverage, but there is no single definitive administrative data source for this coverage type. Interstudy (owned by Clarivate) surveys insurers.
Figure 2 shows estimates of the number of persons enrolled in group health plans. Both
The 2020 MEPS-IC found that the number of private-sector workers enrolled in employer coverage decreased modestly from 62.5 to 60.8 million from 2019 to 2020, a 2.7 percent decline that was not statistically significant./48
These estimates of those with employer coverage are lower than those from Mark Farrah and Interstudy above because they only include workers, not dependents, and do not include those with retiree coverage. Among small employers (<50 employees), the 2020 MEPS-IC found that enrollment in employer coverage declined from 9.6 to 7.9 million due to the reduction in employment among small employers. Among large employers (100+ employees), enrollment in employer coverages was steady from 2019 to 2020.
* * *
[See link at end of text for Figure 2: Group Insurance Enrollment Trends 2019-2021]
Sources: Interstudy (includes off-Marketplace individual insurance)/44 and
* * *
Information on employer coverage can also be gleaned from financial filings to the
* * *
[See link at end of text for Table 3: Insurer Membership Reported in Financial Filings]
Source: ASPE analysis of insurer 10-K filings
* * *
Many adults lost jobs or were furloughed during the pandemic, but did not lose their employer coverage. A
Similarly, while the
Overall, the combined evidence from a variety of data sources to date indicates that the number of uninsured people in the
Administrative data and information on employer coverage show that the moderate decline in employer coverage during the recession noted in several data sources (approximately 2-3 million or 1-3 percent, depending on the data set) was counterbalanced by the greater than 10 million increase in Medicaid enrollment due to federal continuous coverage requirements and to a lesser extent Marketplace growth in 2020.
The more substantial SEP enrollment in the spring and summer of 2021, including the implementation of the ARP's expanded Marketplace subsidies, as well as ongoing growth in Medicaid enrollment, suggests that the uninsured rate may be lower than the pre-pandemic rate; this can be determined once definitive data for 2021 are available. Federal policy actions to maintain or create coverage options for those who might otherwise become uninsured and the rapid economic recovery appear to have tempered the impacts of the pandemic on the total uninsured population. Future federal survey data - ideally after the resolution of 2020's pandemic-related data quality issues - will be essential to clarify these trends.
While overall uninsured rates appear to be largely stable in 2020 and early 2021, federal survey data still indicates sizable disparities in who lacks health insurance coverage. Latinos and Blacks, people with low incomes, and those living in non-expansion states continue to experience much higher uninsured rates than Whites, higher-income households, and those in expansion states.
The information in this report can be used to inform ongoing efforts to expand coverage to the remaining population of approximately 30 million people without health insurance. Increased funding for outreach to the uninsured, enhanced subsidies from the ARP, and the launch of the Marketplace Open Enrollment Period for 2022 all provide important tools to increase the uptake of affordable health insurance coverage and to close disparities in coverage rates.
* * *
View figures, tables, appendix and footnotes at https://aspe.hhs.gov/sites/default/files/documents/2fb03bb1527d26e3f270c65e2bfffc3a/tracking-insurance-coverage-2020-2021.pdf