What to Know About Next Week’s Poverty, Income, and Health Insurance Figures for 2021

On September 13, the Census Bureau will release 2021 poverty, income, and health insurance data from its Current Population Survey (CPS). The figures will show the impact of extraordinary government efforts to bolster economic security and health coverage in the second year of the COVID-19 pandemic, including new data on the reduction in child poverty due to the expanded Child Tax Credit. The figures will reflect the recovering economy in 2021, although not this year’s continued job growth, the expiration of key COVID relief provisions, and higher inflation experienced in much of 2022.

For technical reasons, the release’s “supplemental” poverty indicator will understate the year-over-year improvement in poverty by several million people.

Here are five things to keep in mind.

1. Government Action Protected Very Large Numbers of People From Poverty; Supplemental Poverty Figures Will Show This Best

The Census data will show how many people government support kept above the poverty line in 2021. The figures are likely to make clear that government efforts in the COVID-19 pandemic protected unprecedented numbers of people from poverty in both 2020 and 2021. Last year’s official poverty figures, for example, showed that counting unemployment insurance (UI) lifted the family incomes of a record-high 4.7 million people above the poverty line in 2020, in data back to 1981,[1] chiefly because temporary federal expansions added to program eligibility, duration of benefits, and size of weekly benefits.

Census will also show the poverty reduction in 2021 using the more comprehensive of its two poverty measures, the Supplemental Poverty Measure (SPM). Analysts usually favor the SPM for gauging the poverty-reducing impacts of government programs because, unlike the official (cash-only) poverty measure, the SPM includes the effects on family resources of government non-cash benefits (such as food and housing aid), taxes, and tax credits.

The SPM will account for the American Rescue Plan’s expansion of the Child Tax Credit, which made the full credit available (for 2021 only) to children whose families would otherwise receive a partial credit or no credit at all because their incomes are too low. It also increased the maximum credit (from $2,000 to $3,600 for children under 6 and $3,000 for older children), included 17-year-olds for the first time, and made up to half the credit payable monthly.[2] The SPM figures on people kept above the poverty line by government benefits will also include the impacts of other pandemic relief measures, such as stimulus payments and increased SNAP benefits.

Census will release SPM poverty-reduction data back to 2009. A subsequent CBPP analysis will compare the figures back to 1967 with the help of historical data from Columbia University. A similar CBPP analysis of last year’s SPM data revealed that government programs boosted by COVID relief lifted a record 50 million people above the poverty line in 2020, produced the lowest SPM poverty rate since 1967, and transformed what would otherwise have been the second largest poverty increase on record (not counting government benefits) into the largest poverty drop in over 50 years (when benefits are counted).[3] In 2021, the Child Tax Credit expansion could help lower the SPM child poverty rate further to a new record low, though data challenges (discussed below) could dampen the decline as measured by Census.

2. Health Coverage Likely Improved Again in 2021, Aided by COVID Relief

Next Tuesday’s figures will likely show significant declines in the share of people without health insurance. Data already released from the National Health Interview Survey (NHIS) indicate that the uninsured rate fell in both 2020 and 2021 and could be approaching the record low rates that occurred in the years immediately following implementation of the Affordable Care Act (ACA).[4]

COVID relief legislation was largely responsible for the declining uninsured rate in 2021, evidence indicates.[5] The Medicaid continuous coverage provision, included in the Families First Coronavirus Response Act enacted in the spring of 2020, required that states keep people enrolled in Medicaid for the duration of the public health emergency. As a result, Medicaid and Children’s Health Insurance Program enrollment increased from 71.2 million in December 2020 to 86.7 million by December 2021.[6] (The public health emergency designation is expected to continue until at least January 2023.) The Medicaid continuous coverage provision is also likely why the uninsured rate fell most in 2021 for those with incomes under 100 percent of the federal poverty level, according to the NHIS data.

In addition, five states implemented ACA Medicaid expansions in 2020 and 2021, adding to increases in Medicaid enrollment and declines in uninsured rates for those with low incomes. Oklahoma and Missouri expanded Medicaid in 2021, joining Idaho, Nebraska, and Utah, which expanded in 2020.

Meanwhile, enhanced ACA marketplace premium tax credits and pandemic-related special enrollment periods led to a record 12.2 million people with marketplace coverage in August 2021, up from 10.6 million in August 2020.[7]

The figures Census will release next Tuesday come from the CPS. On September 15, the Census Bureau is scheduled to release estimates from its American Community Survey (ACS), which will include state-level estimates in addition to national estimates. ACS-based comparisons to 2020 will not be possible because pandemic-related data collection disruptions led Census to cancel its standard release of 2020 ACS data. However, comparisons to 2019 ACS data will be possible and will also likely show significant declines in the uninsured rate.

The ACS is larger than the CPS and has maintained a more consistent methodology for health insurance data over the last decade, making it more statistically reliable over time. Looking at results of the ACS in addition to the CPS release will provide more insight into the uninsured rate and its change over time.

3. Supplemental Measure Will Understate the Drop in Poverty by Several Million People, a Close Look at Timing of Stimulus Payments Suggests

Readers should use particular caution when interpreting the SPM data, which will understate the drop in poverty between 2020 and 2021.

Most U.S. residents received $600-per-person stimulus payments in a second round of COVID Economic Impact Payments (EIP2s). (Economic Impact Payments are not included in the official poverty statistics because they are a tax credit and the impact of tax policy is not captured in the official poverty measure, but they are included in the SPM.) Counting the approximately $141 billion in payments when EIP2s were made — overwhelmingly in the first two months of 2021[8] — would make the decline in SPM poverty between 2020 and 2021 much larger. But Census instead counted these payments as 2020 income, following long-standing Census practice to count tax-based benefits in the relevant tax year.[9]

While career Census staff made the decision on sound technical grounds,[10] the decision will greatly affect comparisons of poverty in 2020 and 2021. We calculate that EIP2s alone lifted 3.8 million individuals from poverty in 2020 using the SPM. (The first and second rounds of pandemic stimulus payments together lifted 11.7 million people above the SPM poverty line, last year’s Census report showed.) If Census had counted EIP2s in 2021, rather than in 2020, SPM poverty would have been 3.8 million people (about 1.2 percentage points) higher in 2020 and a similar amount lower in 2021.[11]

As a result, poverty’s decline from 2020 to 2021 could be as much as about 7 million people (2.3 percentage points) larger than the new Census data will show.

4. Income and Poverty Comparisons Could Understate (or Overstate) Recent Progress for Other Technical Reasons; Census Will Assess Impact

The pandemic’s effects on data collection in recent years (on top of the EIP2 issue discussed above) could also skew comparisons to earlier years’ income and poverty data in either direction. Census will release an assessment of the impact of these pandemic-related data problems on the topline CPS income figures.

A similar Census assessment last year noted that fewer people completed the survey than usual in the pandemic, particularly those with low incomes; it concluded that Census’s own median household income figures were likely about 2.0 percent too high in 2020 and 2.9 percent too high in 2019, as a result.[12] Other data Census released this February suggest that the CPS official poverty rate for 2020 was about 0.4 percentage points too low due to data quality issues.[13]

Users should pay close attention to income and poverty figures Census includes in its quality assessment. Further, if the assessment finds that pandemic-related inaccuracies in the official poverty measure have largely eased in the latest data, this could mean that the latest CPS poverty findings are more comparable with poverty figures for the pre-pandemic year of 2018 than with figures for 2019 or 2020. In this case, users may wish to focus on how the 2021 poverty data compare with 2018.[14]

5. Lower Unemployment Benefits Could Offset Employment Gains, Leaving Income and Official Poverty Rate Little Changed From 2020

The recovering labor market in 2021 could lead to modest progress on annual measures of income and official poverty status (where only cash income before taxes is considered) in the pandemic’s second year. Employers added 3.9 million jobs (comparing annual averages in 2020 and 2021), reversing nearly half the job loss from 2019 to 2020. Average weekly earnings of private non-supervisory workers edged up 0.5 percent after adjusting for inflation, according to the Labor Department, while another wage indicator, median usual weekly wages of full-time workers, declined 3.2 percent.

These mostly positive influences on income could be largely offset by the year’s sizable decline in unemployment insurance payments, however, and could be further dampened in the Census data by the pandemic’s effects on data quality noted above.

End Notes

[1] U.S. Census Bureau, “Impact on Poverty of Alternative Resource Measures by Age: 1981 to 2020,” https://www2.census.gov/programs-surveys/demo/tables/p60/273/Impact_Poverty.xlsx.

[2] Census is unlikely to distinguish between the advance part of the Child Tax Credit that Treasury delivered monthly — up to $300 per child per month between July and December of 2021 — and the balance that IRS paid as refunds after families filed their 2020 tax forms.

[3] The CBPP analysis merged Census SPM data with historical SPM data produced by Columbia University researchers. The basic findings — a record-low SPM poverty in 2020, and the transformation of a near-record poverty increase (measured without government assistance income) into the largest poverty reduction since 1968 (when government assistance is included) — hold up even after accounting for two technical issues raised below regarding the timing of stimulus payments and other survey problems in the pandemic.

[4] The NHIS data indicate that the uninsured rate declined in 2021, from 9.5 percent in the first quarter to 8.8 percent in the fourth quarter. For comparison, the NHIS full-year uninsured rate was 9.7 percent in 2020 and 10.3 percent in 2019, and the lowest full-year uninsured rate ever recorded by the NHIS was 9.0 percent in 2016. Moreover, the uninsured rate reached a quarterly record low of 8.0 percent in the first quarter of 2022, according to the NHIS, though this is outside the 2021 timeframe of the forthcoming Census data. Centers for Disease Control and Prevention, “Health Insurance Coverage: Estimates from the National Health Interview Survey,” https://www.cdc.gov/nchs/nhis/healthinsurancecoverage.htm. Robin Cohen and Amy Cha, “Health Insurance Coverage: Early Release of Quarterly Estimates From the National Health Interview Survey, January 2021–March 2022,” Centers for Disease Control and Prevention, July 2022, https://www.cdc.gov/nchs/data/nhis/earlyrelease/Quarterly_Estimates_2022_Q11.pdf.

[5] Gideon Lukens, Jennifer Sullivan, and Farah Erzouki, “COVID Relief Provisions Stabilized Health Coverage, Improved Access and Affordability,” CBPP, March 10, 2022, https://www.cbpp.org/research/health/covid-relief-provisions-stabilized-health-coverage-improved-access-and.

[6] Centers for Medicare & Medicaid Services, Monthly Medicaid and CHIP Application, Eligibility Determination, and Enrollment Reports & Data, https://www.medicaid.gov/medicaid/national-medicaid-chip-program-information/medicaid-chip-enrollment-data/monthly-medicaid-chip-application-eligibility-determination-and-enrollment-reports-data/index.html.

[7] Centers for Medicare & Medicaid Services, “Total Effectuated Enrollment by State, August 2019-2021,” https://www.cms.gov/files/document/2019-2021-aug-effectuated-enrollment.xlsx. These enrollment statistics, known as “effectuated enrollment,” measure the number of people who are actively enrolled in the ACA marketplaces in August and have paid their monthly premium. Another statistic commonly cited, open enrollment period plan selections, includes people who sign up for marketplace coverage but do not necessarily effectuate their coverage by paying premiums. While effectuated enrollment best captures the number of people who are actively enrolled in the marketplaces, statistics for plan selections are often cited because they are known well in advance of effectuated enrollment data. For example, while 14.5 million people selected marketplace plans during 2022 open enrollment, data for effectuated enrollment in 2022 have not yet been released.

[8] Enacted on December 27, 2020, at least 99 percent of EIP2 dollars were paid in 2021, monthly Treasury figures show.

[9] The $600 stimulus payments were technically advance payments of a 2020 tax credit and were administered by IRS. For background and funding levels for all three stimulus payments, see https://www.irs.gov/statistics/soi-tax-stats-coronavirus-aid-relief-and-economic-security-act-cares-act-statistics.

[10] The CPS does not ask families directly about taxes or tax credits; instead, Census estimates them for families, which avoids adding needlessly to an already long survey. Census bases its estimates on respondents’ income in the legally relevant tax year, ensuring that the estimated value of the credits and the demographic profile of the assigned recipients are as accurate as possible.

EIP2s differ somewhat from other tax credits counted by Census in that their late-December enactment made it impossible for most families to anticipate their arrival during the year. Tax filers typically only receive refundable tax credits the following year after filing taxes. But in the case of most credits, families may be able to anticipate receiving the credit and reduce their tax withholding accordingly.

[11] The number of people EIP2s removed from poverty might be somewhat different if counted in 2021 than in 2020 (most likely smaller) because of differences such as the improving economy.

[12] CPS income and poverty data for 2019 were gathered in February, March, and April 2020. The Census data quality assessment was informed in part by employer pay records concerning CPS households that did and did not complete the survey in each of the last five years. Jonathan Rothbaum and Charles Hokayem, “How Did the Pandemic Affect Survey Response: Using Administrative Data to Evaluate Nonresponse in the 2021 Current Population Survey Annual Social and Economic Supplement,” U.S. Census Bureau, September 14, 2021, https://www.census.gov/newsroom/blogs/research-matters/2021/09/pandemic-affect-survey-response.html.

[13] In addition, the CPS poverty rate for 2019 was about 0.5 percentage points too low. CBPP analysis of Census CPS data at https://www.census.gov/data/datasets/2020/demo/cps/2020-pub-use-adj-nonresp-corono-pandemic.html.

[14] Due to a separate issue, it will be easier to accurately compare 2018 with 2021 in the case of CPS poverty data than in the case of median income. As it does every ten years, Census is switching from basing the CPS on the 2010 decennial census to basing it on the 2020 decennial census. This switch makes little difference — no more than 0.1 percentage point — for poverty rates overall, by age group, or for Black, Latino, white, or Asian populations. But the change matters more for income: it raises median household income in 2020 by a substantial 0.72 percent, a Census Bureau paper finds. (Census produced CPS data for 2020 each way, based on both the 2010 and 2020 censuses, but will produce 2021 data based only on the 2020 census.) To compare median income from 2018 and 2021, users should increase 2018 income by 0.72 percent. See Em Shrider, Jessica Semega, and Katherine K. Starkey, “Effects of 2020 Census-Based Population Controls on 2020 Income, Poverty, Supplemental Poverty, and Health Insurance in the United States Estimates,” U.S. Census Bureau working paper, August 2022, https://www.census.gov/content/dam/Census/library/working-papers/2022/demo/sehsd-wp2022-14.pdf.