Blog & News
Current Population Survey (CPS) will Serve as Primary Source of 2020 State-level Data on Health InsuranceSeptember 22, 2021:
|RECENT CPS ANALYSIS|
On September 14, the U.S. Census Bureau released 2020 health insurance estimates from the Current Population Survey (CPS). These data will serve as one of the only sources of 2020 state-level health insurance as the Census Bureau will not be releasing its typical 1-year estimates from the 2020 American Community Survey (ACS) due to impacts of the COVID pandemic that resulted in substantially lower response rates and nonresponse bias.
SHADAC typically relies on the ACS to study state and sub-state (e.g., state coverage by race) health insurance trends and posts detailed state estimates on State Health Compare. However, because the ACS data are not being released this year, we recommend that analysts instead use the CPS and have posted 2020 estimates from the CPS on State Health Compare for analysts and policymakers that need 2020 state-level information on coverage. This blog provides an overview of important differences between the two surveys for those using CPS estimates in place of the ACS this year.
Key differences between CPS and ACS
There are some critical differences between the CPS and ACS, and it is important to understand these differences when interpreting results.
Perhaps one of the most critical differences between the ACS and the CPS for analysts and policymakers interested in estimates at the state, and more granular levels of geography or subpopulations, is sample size. The ACS has significantly more sample than the CPS; over 2 million households in 2019 compared to just over 94,000 in the CPS. Sample size is the primary reason that SHADAC typically uses the ACS to produce estimates of coverage for State Health Compare. As we note in our blog outlining results from the CPS, relying on the CPS means that we are unable to produce as many subpopulation estimates within states as we do with the ACS. In addition to limiting the estimates that can be produced, the smaller sample size of the CPS also results in less precision, even for estimates that the sample size does support.
Conceptual differences in the definition of the uninsured
The CPS estimates presented above and on State Health Compare reflect insurance status of respondents for the entire calendar year of 2020. In contrast, the ACS collects information at a “point in time” when the survey was conducted. As a result, estimates of uninsurance in the CPS are lower than the ACS, because people who had coverage at some point in the prior calendar year are not considered uninsured. The ACS, on the other hand, captures a cross-section of people who are uninsured at the time the survey was conducted, some of whom were not uninsured for the entire year. These differences are important, but it is also helpful to note that the surveys have historically demonstrated similar national trends over time, and the patterns across states are consistent in that states with low uninsurance levels have low levels in both surveys, and states with high levels have high levels in both surveys, and so on. For more detailed comparisons across surveys in prior years, see SHADAC’s “Comparing Federal Government Surveys That Count the Uninsured” brief.
As discussed above, the CPS asks about coverage for the entire previous calendar year. The survey is fielded in February through April of each year, with respondents being asked to report their coverage for a time period as long as 16 months prior to the interview. The ACS collects information about current coverage only. Differences in the length of time for which respondents are being asked to recall their insurance coverage status can result in differences in measurement error across the surveys.
Breadth of Related Content
Another important difference between the two surveys is the breadth of related content. The ACS’ main focus is broad demographic information, with just one question on health insurance. While the rich demographic information supports examination of the uninsured and the exploration of coverage by individual characteristics (e.g., social determinants of health) it has limited health policy applications. The CPS, on the other hand, contains a range of measures that are broadly relevant to health policy, such as medical out of pocket spending, health status, and eligibility for employer coverage. This means that the survey can be used to answer more complex research questions about the interactions between coverage and these related outcomes than could be supported through analysis of the ACS. The CPS also collects much more detailed information on income and employment than the CPS.
The CPS provides a good alternative for those that usually rely on state-level estimates of coverage from the ACS for analysis and planning. However, the data source does have some limitations, particularly for those seeking more granular estimates by geography and subpopulations within states. SHADAC is available to provide technical assistance to states that are seeking additional guidance on potential data sources to leverage for answering these questions in 2020. We will also be tracking the Census Bureau’s release of the 2020 1-year experimental ACS data, which are expected by November 30 of this year.
U.S. Census Bureau. (2016, May 16). Fact sheet: differences between the American Community Survey (ACS) and the Annual Social and Economic Supplement to the Current Population Survey (CPS ASEC). https://www.census.gov/topics/income-poverty/guidance/data-sources/acs-vs-cps.html
State Health Access Data Assistance Center (SHADAC). (October 2020). Comparing federal government surveys that count the uninsured: 2020. https://www.shadac.org/publications/comparing-federal-government-surveys-count-uninsured-2020
Stewart, A. (2021, July 30). Census Bureau announces major changes to 2020 American Community Survey (ACS) data release. State Health Access Data Assistance Center (SHADAC). https://www.shadac.org/news/changes-to-2020-acs-data-release-US-Census
Stewart, A. (2021, August 27). New SHADAC brief looks at changes in federal surveys during COVID pandemic. State Health Access Data Assistance Center (SHADAC). https://www.shadac.org/news/new-shadac-brief-looks-changes-federal-surveys-during-covid-pandemic
Blog & News
Current Population Survey Shows 2020 National Uninsured Rate Stable, Rising in Three StatesSeptember 22, 2021:
On September 14, the U.S. Census Bureau released 2020 national health insurance estimates from the Current Population Survey along with public use microdata. These data will serve as one of very few sources of information on 2020 state-level health insurance as the U.S. Census Bureau will not release its normal set of 1-year estimates from the 2020 American Community Survey (ACS), due to impacts of the pandemic that resulted in nonresponse bias and substantially lower response rates.
Given its large sample size, SHADAC typically relies on the ACS to study state and sub-state (e.g., county-level or state-level coverage by race) health insurance trends and posts detailed state estimates on State Health Compare. However, because the ACS data are not being released as usual this year, we recommend that analysts use the CPS, and have posted 2020 estimates from this survey on State Health Compare for analysts and policymakers that need 2020 state-level information on coverage. Differences between the ACS and CPS and considerations for their use are summarized here.
This post presents highlights from the 2020 state-level coverage estimates on State Health Compare and compares 2020 estimates to 2018, a pre-COVID baseline unaffected by pandemic-related data collection challenges.
Uninsurance was stable nationally and in most states
In 2020, 8.6 percent of Americans (nearly 28 million people) were uninsured all year, statistically unchanged from a pre-pandemic baseline of 8.5 percent in 2018. Rates of uninsurance were unchanged in most states, though three states (Arizona, Missouri, and Tennessee) saw increases and five states (Florida, Maryland, Oregon, Vermont, and Virginia) experienced decreases. Tennessee had the largest increase at 4.1 percentage points (PP) (11.4 percent vs. 7.3 percent), and Virginia had the largest decrease at 3.3 PP (5.5 percent vs. 8.8 percent).
More Americans had public coverage and fewer had private coverage
The percent of Americans with public coverage at some point during 2020 increased to 32.8 percent from 32.3 percent in 2018. This equates to 2.4 million more people with public coverage at some point in 2020 as compared to 2018. At the state level, seven states had increases in rates of private coverage (Maryland, Massachusetts, Michigan, New Hampshire, Ohio, Oklahoma, and Wyoming), and only Virginia had a decrease in rates of private coverage. Of these states, Maryland had the largest increase in public coverage at 6.2 PP (31.2 percent vs. 25.0 percent) and Virginia had the largest and only decrease in public coverage at 3.4 PP (25.1 percent vs. 28.5 percent).
The percent of Americans with private coverage at some point during 2020 fell to 58.6 percent from 59.2 percent in 2018, which represents 934,000 fewer people with private coverage. In most states, however, the percent with private coverage remained stable. Just two states (Virginia and West Virginia) saw increases in private coverage, and five states (Colorado, Delaware, Massachusetts, Ohio and Tennessee) saw decreases. Of these states, Virginia had the largest increase at 6.7 PP (69.5 percent vs. 62.7 percent), and Tennessee had the largest decrease at 7.1 PP (54.7 percent vs. 61.7 percent).
2020 coverage estimates available on State Health Compare
In addition to state-level estimates by both broad coverage types (Insured, Private, Public, and Uninsured) and more detailed (Employer, Individual, Medicaid/CHIP, Medicare, and Uninsured), state-level coverage estimates are available by age group (0-18, 0-64, 19-64, and 65+) and health status (Good/Very Good/Excellent and Fair/Poor). Estimates by poverty level will likely be forthcoming. However, coverage estimates by race/ethnicity for most states do not meet our standards for statistical reliability and precision due to the relatively small sample size of the CPS.
Data users should also note that the 2020 State Health Compare coverage estimates from the CPS are not comparable to estimates from the ACS, since the two surveys use different concepts of health insurance coverage and uninsurance. The CPS asks respondents if they had a particular type of coverage at any point during the previous year or if they were uninsured for the entire year. The ACS asks respondents about their health insurance coverage at the time of the interview. More information on this topic will be available in a forthcoming related brief.
Blog & News
2021 CPS ASEC: Fewer Americans Had Health Insurance Coverage in 2020 — Private Coverage Fell while Public Coverage RoseSeptember 15, 2021:
The U.S. Census Bureau has released 1-year health insurance coverage estimates for 2020 from the Current Population Survey Annual Social and Economic Supplement (CPS ASEC). The Census report, “Health Insurance Coverage in the United States: 2020,” showed that the national uninsured rate was 8.6 percent in 2020, which represented a significant increase from 8.0 percent in 2019.1
No state-level estimates were published in the report, which in the past has drawn on data from the Census Bureau’s American Community Survey (ACS) for information on health insurance coverage by state. Estimates from the 2020 1-Year ACS will not be released as usual this year due to substantial COVID-19-pandemic-related disruptions to data collection operations and resulting issues with data quality. However, the CPS ASEC does provide important breakdowns of available health insurance coverage estimates at the national level, which are explored in this blog, and SHADAC plans to soon release state-level 2020 health insurance coverage estimates from the CPS ASEC.
In 2020, 66.5 percent of the United States population had private coverage, which represented a 1.6 percentage-point (PP) decrease from 68.0 percent in 2019. Conversely, the portion of the U.S. population with a public plan increased by 0.7PP to 34.8 percent in 2020 from 34.1 percent in 2019.
Among categories of private-based insurance coverage…
… employer-sponsored insurance (ESI) continued to be the most common form of coverage at 54.4 percent, followed by direct-purchase coverage (including marketplace coverage) at 10.5 percent, individual marketplace coverage at 3.3 percent, and TRICARE at 2.8 percent.
Among categories of public-based insurance coverage…
… Medicare and Medicaid were split nearly evenly at 18.4 percent and 17.8 percent, respectively. Individuals who had the Department of Veterans Affairs (VA) or Civilian Health and Medical Program of the Department of Veterans Affairs (CHAMPVA) made up 0.9 percent of public coverage.
Medicaid Expansion Status
Differences in uninsured rates were also apparent when examined by state Medicaid expansion status. In 2020, the rate of uninsured for states that expanded Medicaid was 6.4 percent, and in non-expansion states the uninsured rate for 2020 measured at 12.6 percent.
Looking at rates of uninsured individuals by age, children age 19 and younger measured at 5.6 percent in 2020. Adults under age 65 had an uninsured rate of 10.2 percent in 2020, while uninsured rates for adults over the age of 65 measured at just 1.0 percent in 2020.
Estimates of uninsurance by race/ethnicity were also included within the CPS ASEC report. For 2020, the portion of the population who identified as White had an uninsured rate of 8.3 percent, those who identified as Black had an uninsured rate of 10.4 percent, the uninsured rate for those who identified as Asian was 5.9 percent, and those who identified as Hispanic regardless of their race had a rate of 18.3 percent who were uninsured.
The report also provided health insurance coverage data by poverty levels. The rate of uninsured for 2020 for those below 100 percent of poverty measured at 17.2 percent and individuals below 138 percent of poverty saw a similar rate of 16.1 percent in 2020. Conversely, individuals at or above 400 percent of poverty had an uninsured rate of 3.4 percent.
Current and Future Data Releases and Products
In addition to their health insurance coverage report, the Census Bureau also published two new companion blogs to the CPS ASEC release.
The first post discusses the way in which the CPS has changed its measure of health insurance coverage within a particular year as a result of the 2014 questionnaire redesign, as well as providing an overview of subannual coverage in the CPS from 2018 to 2020, and whether and how these measures may be compared with estimates from other surveys such as the Medical Expenditure Panel Survey (MEPS) and the Survey of Income and Program Participation (SIPP).
The second blog takes a deeper dive into the effects of the COVID pandemic on response rates for the 2021 CPS ASEC and the extent that nonresponse bias played a factor in the outcome of the 2020 estimates for income, poverty, and health insurance coverage.
Experimental data from the 2020 ACS 1-Year Estimates are expected to be published by the Census Bureau in November 2021, and 2016-2020 ACS 5-Year Estimates are planned for release in December 2021 (though Census is still evaluating whether the 5-Year ACS estimates meet their standards for data quality).
SHADAC will monitor these planned ACS releases and will assess whether and how any of these products could be used to produce state-level health insurance coverage and related estimates. We will also assess if and which other data sources may be useful for producing the coverage estimates we typically get from the ACS.
1 All statistically significant differences in this post are evaluated at the 90% confidence level.
2019 estimates from the CPS ASEC were impacted by disruptions due to the COVID-19 pandemic, which likely resulted in measurable nonresponse bias. Comparisons with these 2019 estimates should be treated with caution. See Census Working Paper SEHSD WP2020-10 for more information.
Impacts of the Affordable Care Act Medicaid Expansion in California (CJPP)
This Journal article was originally published in the California Journal of Politics and Policy.
A cornerstone aim of the Affordable Care Act (ACA) was to make health insurance coverage readily available to more people in the United States through policies such as new federal subsidies to help people with moderate incomes purchase private insurance through new Marketplaces and expanding Medicaid’s reach to many more low-income adults. Key components of the ACA were implemented in 2014, including state-level expansions of Medicaid. Studies that compared states that opted for and against expanding their Medicaid programs have documented gains in coverage, access, and health status particularly among low-income and childless adults, who were the main subpopulation beneficiaries of Medicaid expansion in states that adopted it.i However, few of these studies have focused specifically on California. To our knowledge, past studies on the effects of Medicaid expansion in California have not examined self-reported general health and physical/mental health for low-income adults. Self-reported health measures are important for monitoring population health over time, quantifying individual satisfaction with health, and complementing objective long-term outcomes such as mortality.ii
This journal article authored by SHADAC researchers Natalie Schwehr, Giovann Alarcón, and Lacey Hartman examines the impact of the Affordable Care Act (ACA) on healthcare coverage, access, health status, and affordability, as well as disparities in these outcomes by race/ethnicity among low-income Californians.
Using nationally representative survey data from the Behavioral Risk Factor Surveillance Survey 2011-2019 and a difference-in-differences approach that compared California with nonexpansion states, they examined the impact of Medicaid expansion on health insurance coverage, having a usual source of care, self-reported health status, frequent (≥14) unhealthy days in the past month (physical, mental, and both), and foregone care due to cost.
The sample population included low-income Californians (<100% of the federal poverty guidelines) aged19-64 and low-income childless adults.
- Low-income adults, childless adults, and white childless adults in California saw post-ACA gains in six of seven outcomes, including a 7.7 percentage point increase in having a usual source of care for all low-income adults.
- Childless adult people of color (POC) reported significant improvements in three measures, with a 6.6 percentage point increase in having a usual source of care (CI: 0.013 to 0.120).
- All groups examined had coverage gains, ranging from 3.9 percentage points for all low-income adults, to 8.4 percentage points for white childless adults.
- All groups reported improved mental health, including an 8.2 percentage point decrease in frequent mental distress for childless adults.
These findings indicate that the ACA coverage expansion benefitted the targeted population of low-income Californians. Additionally, the disparity between white and non-white Californians decreased for the unadjusted mean rate of having a usual source of care. However, unadjusted means showed that white low-income adults remained more likely to have health insurance coverage and a usual source of care compared with POC in both California and nonexpansion states.
Schwehr, N., Alarcón, G., & Hartman, L. (2021). Impacts of the Affordable Care Act Medicaid Expansion in California. California Journal of Politics and Policy, 13(1). http://dx.doi.org/10.5070/P2cjpp13154573 Retrieved from https://escholarship.org/uc/item/17d520j7
i Simon, Kosali, Aparna Soni, and John Cawley. 2017. “The Impact of Health Insurance on Preventive Care and Health Behaviors: Evidence from the First Two Years of the ACA Medicaid Expansions.” Journal of Policy Analysis and Management36 (2): 390–417. https://doi.org/10.1002/pam.21972