Poverty During the Pandemic and the Role of Government Transfers

The COVID-19 pandemic prompted unprecedented policy interventions in the US that provided nearly $3 trillion to support struggling families. This column examines the short-term effects of these interventions on child poverty and finds quite different trends for income poverty and consumption poverty. While disposable income poverty declined dramatically in 2021, consumption poverty fell more gradually, highlighting the role of saving and borrowing in shaping economic outcomes for low-income families. Income poverty rose in 2022 but consumption poverty continued to fall and was noticeably lower than prior to the pandemic, indicating that pandemic-era policies played an important role in maintaining economic wellbeing during this period.

Policymakers have long debated how best to alleviate child poverty and what role government support should play. The COVID-19 pandemic, which led to widespread economic disruption, prompted unprecedented policy interventions that provided nearly $3 trillion to support struggling families. These measures included Economic Impact Payments (EIPs), or stimulus payments, and expansions in the generosity and eligibility of unemployment insurance and the Child Tax Credit (CTC). Understanding the short-term effects of these interventions on child poverty is crucial, especially as policymakers continue to debate whether and how to extend or renew such measures in the future.

The US Census Bureau, and several research studies, have examined income-based poverty estimates before and after the COVID-19 pandemic and considered how certain policies have affected these estimates (Shrider and Creamer 2023). However, there are key reasons to re-examine these income-based statistics and consider other measures of economic wellbeing. In particular, due to misreporting, the income measures fail to capture much of the income support provided by the government. The misreporting of government transfers in surveys is well-documented (Meyer et al. 2015), but the extent of this misreporting grew significantly as these transfers expanded during the pandemic. For example, income data in the Current Population Survey, which is the source of official poverty statistics, captured less than 40% of total unemployment insurance benefits in 2020, missing more than $360 billion in income, the vast majority of which went to low-income individuals and families. There are other reasons why income-based poverty statistics might not accurately reflect changes in economic wellbeing during the pandemic. For example, uncertainty about future earnings as a result of the large and unprecedented economic disruption from the pandemic may have led some individuals to save part of their temporary government transfers.

To assess changes in economic wellbeing during the pandemic, we consider two different poverty measures: one based on income and one based on consumption. We look at ‘disposable income’, which combines after-tax income with the value of non-cash benefits, such as food subsidies from the Supplemental Nutrition Assistance Program (SNAP) and housing subsidies. This measure, which is constructed using data from the Current Population Survey, includes the income from the major policies aimed at improving the economic circumstances of individuals and families during the pandemic, including the Economic Impact Payments (EIPs or stimulus payments), the Child Tax Credit (CTC), and unemployment insurance.

Our measure of consumption is constructed using data from the Consumer Expenditure Survey. It includes spending on goods and services, as well as the value of services that families derive from durable goods like housing and vehicles (see Meyer et al. 2024 for details on how these measures are constructed). We examine these income- and consumption-based poverty estimates from 2015 through 2022, a period that starts several years before the pandemic and extends at least one year after the largest pandemic policies had expired.

Figure 1 Consumption and income child poverty rates, 2015-2022

Notes: Thresholds anchored in 2015.Official Income Poverty follows the US Census definition of income poverty using official thresholds. For measures other than the official one, the 2015 threshold is set to the value that produces a poverty rate equal to the official 2015 poverty rate of 13.5% for the full sample. These thresholds are then adjusted over time using the Bias-Corrected CPI-U-RS, which subtracts 0.8 percentage points from the growth in the CPI-U-RS each year from 2015 to 2022. Poverty status is determined at the family level and then person-weighted. Disposable income includes taxes and credits (calculated using TAXSIM) as well as the value of noncash benefits such as SNAP, WIC, school lunch, energy assistance, and housing subsidies. Consumption data are from the CE, and income data are from the CPS-ASEC.
Source: Meyer et al. (2024).

Several factors help explain this divergence. First, changes in saving and borrowing behaviour during the pandemic played a role. Families who received large income transfers, such as stimulus payments, expanded unemployment benefits, and the CTC in 2021 saved part of this extra income, causing income to rise more than consumption. This behaviour is evident in the data on assets for families with few resources. In Figure 2, we report the 75th percentile of assets for families with children and with consumption between 50% and 150% of the poverty line. For this group assets rose sharply in the second quarter of 2020, with the first round of EIPs. Assets rose even more sharply in the second quarter of 2021, a month or two after the second round of EIPs were sent out, and around the same time that the third round of EIPs were sent out, but before the advance CTC payments were made. Assets then returned to pre-pandemic levels in 2022, suggesting that these families were dissaving, allowing families to maintain (or even increase) consumption as income fell. Other studies have shown evidence suggesting that families increased saving in response to pandemic policies, particularly the EIPs (Bachas et al. 2020, Wheat and Deadman 2023).

Figure 2 Seventy-fifth percentile of assets for families with children and total consumption between the 50-150% poverty threshold

Notes: This figure shows the 75th percentile of assets for families with consumption between 50 and 150 percent of the poverty threshold, with the 2015Q1 poverty threshold set to the value that produces a poverty rate equal to the official 2015 poverty rate of 13.5 percent for the full sample. These thresholds are then adjusted over time using the Bias-Corrected CPI-U-RS. Total assets are calculated as the sum of the values of all checking, savings, and money market accounts, certificates of deposit (CDs), directly-held stocks, bonds, and mutual funds. The data are from the CE. Asset information is collected only in the fifth interview of the CE. The quarter reflects the survey month for the data.
Source: Meyer et al. (2024).

Measurement error is another potential explanation for why income and consumption poverty diverge during the pandemic. Previous research has shown that government income transfers are significantly under-reported and this under-reporting has grown over time (Meyer et al. 2015). Moreover, as noted above, unemployment insurance benefits were significantly under-reported during the pandemic. We try to account for this under-reporting by imputing benefits to some survey respondents who did not report receipt so that the total amounts match those reported in administrative records. We also impute the values of the EIPs and CTC to match these administrative totals. Nevertheless, errors in this imputation process, as well as changes in under-reporting of other programmes, might lead to differences between these two measures of poverty. Another possibility is under-reported private transfers which may have replaced a share of changes in public transfers.

How COVID-era policies affected child poverty

All three of the major pandemic income support programmes – EIPs, enhanced unemployment insurance, and the expanded CTC – played an important role in the fall in child poverty during the pandemic. One way to assess the effect that each of these programmes had on income poverty is to estimate income poverty both including and excluding each policy.¹For example, in 2021, without the expanded CTC, child poverty would have decreased by just 2.9 percentage points, compared to the actual 3.9 percentage point reduction. This suggests that, without the expanded CTC, child poverty would have been 1 percentage point higher in 2021 (Figure 3). Similarly, child poverty would have been 3.1 percentage points higher without the EIPs and 1.2 percentage points higher without expanded unemployment insurance (UI). Thus, for 2021 the second and third rounds of EIPs had, by far, the largest impact on reducing child income poverty, although both UI and the expanded CTC also played a role. When we conduct similar analyses for 2022, we find that the sharp rise in income poverty in that year is primarily due to the termination of the EIPs, with the expiration of the expanded CTC playing a smaller role. This is partly because, although the expanded CTC counted as 2021 income for tax purposes, about half of this credit was received in 2022. Note that we consistently included the 2020 CTC as having been entirely received in tax refunds in 2021 (as it was in reality).

Figure 3 Change in poverty between 2020 and 2021 relative to baseline change

Notes: This figure shows the relative changes in income poverty of children under the counterfactual policies of replacing the expanded CTC with the original CTC (left bar), excluding EIPs 2 and 3 (middle bar), or excluding UI (Unemployment insurance) (right bar), compared to the baseline change of 3.9 percentage points under the actual policies.
Source: Meyer et al. (2024).

Implications

Our findings provide insights into how child poverty changed during the pandemic and the role of COVID-era policies played in these changes. Our results show that the trends for income poverty during the pandemic were quite different from those for consumption poverty. While poverty based on disposable income showed a dramatic decline in 2021, consumption poverty fell more gradually, highlighting the role of saving and borrowing in shaping economic outcomes for low-income families. Consumption poverty continued to fall in 2022 while income poverty rose. Consumption poverty in 2022 was noticeably lower than it was prior to the start of the pandemic, indicating that pandemic era policies played an important role in maintaining economic well-being during this period. Despite this evidence of improved material circumstances, individuals and families did struggle along other dimensions. Families both with and without children reported higher rates of financial and food hardships in 2022. Increasing hardship during a time when material circumstances improved is consistent with recent studies showing that even though overall economic conditions were improving during this period consumers were becoming increasingly concerned with rising prices and borrowing costs (Bolhuis et al. 2024).

References

Bachas, N, P Ganong, P Noel, J Vavra, A Wong, D Farrell and F Greig (2020), “Initial Impacts of the Pandemic on Consumer Behavior: Evidence from Linked Income, Spending, and Savings Data”, NBER Working Paper No. 27617.

Bolhuis, M A, J Cramer, K O Schulz and L H Summers (2024), “Consumers regard the cost of money as part of the cost of living”, VoxEU.org, 7 May.

Han, J, B D Meyer and J X Sullivan (2020), “Income and Poverty in the COVID-19 Pandemic”, Brookings Papers on Economic Activity.

Meyer, B D, J Han and J X Sullivan (2024), “Poverty, Hardship, and Government Transfers”, NBER Working Paper No. 33052.

Meyer, B D, W K C Mok and J X Sullivan (2015), “Household Surveys in Crisis”, Journal of Economic Perspectives 29(4): 199–226.

Shrider, E A and J Creamer (2023), “Poverty in the United States: 2022”, U.S. Census Bureau, Current Population Reports, P60-280.

Wheat, C and E Deadman (2023), “Household Pulse: Balances through March 2023”, JPMorgan Chase Institute.

Footnotes

For this analysis we count EIPs and the CTC as income in the year they were received, rather than the year they were accrued; see Meyer et al. (2024) for more details.