Child Tax Credit expansions were instrumental in reducing poverty rates to historic lows in 2021

Government policies enacted in the wake of the pandemic have proven critical for reducing child poverty in the United States. Census Bureau data released last week showed that government social programs kept tens of millions of people out of poverty in 2021.

Child poverty reached its lowest level on record, as calculated by the Supplemental Poverty Measure (a measure that includes both cash and noncash benefits). This new historic low is largely thanks to the expanded Child Tax Credit (CTC), a key component of the 2021 American Rescue Plan (ARP) that has since expired. Without additional action by Congress to renew the expanded Child Tax Credit, we should expect higher child poverty in future years.

Let’s start with the outstanding role the Child Tax Credit played in reducing child poverty. The Child Tax Credit is a payment to support families raising children under 17 years of age of up to $2,000 per qualifying child. The 2021 ARP expanded the credit to increase the level of earnings to families receiving the credit (up to $3,600 per child under age 6) and to make the credit more widely available and fully refundable.

The refundable Child Tax Credit alone accounts for a reduction in child poverty of 2.9 million. Within that, the expanded Child Tax Credit—a key element of the 2021 American Rescue Plan (ARP)—lifted 2.1 million children out of poverty. The ARP Child Tax Credit is the leading reason child poverty fell so precipitously from 9.7% in 2020 to 5.2% in 2021, the lowest rate on record. Nearly three-quarters of the poverty-reducing impact of the Child Tax Credit came from the ARP expansions. In total, the increasing importance of the Child Tax Credit is responsible for about 70% or 3.1 percentage points of that 4.5 percentage-point reduction in poverty between 2020 and 2021.

Figure A separates out the effects of the Child Tax Credit without the expansions and the expanded Child Tax Credit on children’s poverty by race and ethnicity. White non-Hispanic child poverty was lower by 820,000 in 2021 because of the Child Tax Credits, 649,000 of which came from the expansions. In addition, 716,000 fewer Black children were in poverty in 2021 because of the Child Tax Credit—over 80% of that reduction in poverty came from the ARP expansions to the Child Tax Credit, one of the key reasons why Black child poverty fell by more than half between 2020 and 2021. Hispanic child poverty also saw dramatic reductions from the expansions.

Figure A

The expanded Child Tax Credit reduced child poverty enormously: Number of children lifted out of poverty by each measure, in thousands

Child Tax Credit (not including the ARP expansions) Expanded Child Tax Credit (only) Total
White non-Hispanic 171 649 0
Black 116 600 0
Asian 54 56 0
Hispanic 428 752 0
ChartData Download data

The data below can be saved or copied directly into Excel.

Notes: Federal surveys give respondents the option of reporting more than one race. In this case the U.S. Census Bureau uses the race-alone concept where, for example, a group such as Asian may be defined as those who reported Asian and no other race. ARP refers to the 2021 American Rescue Plan

Source: EPI analysis of United States Census Bureau Child Tax Credit data. 

Copy the code below to embed this chart on your website.

Some debate has surfaced around whether the child poverty rate was the lowest on record, pertaining to the Supplemental Poverty Measure (SPM) and breaks in the series that not only impact the SPM, but also the official poverty measure, median household income, and other metrics related to the Current Population Social and Economic Supplement (CPS ASEC). The CPS ASEC—sometimes referred to as the March CPS—asks retrospective questions of households in the spring of one year about their economic circumstances in the prior year. Over the last decade, the Census has changed the survey questions and methodology to more accurately measure households’ economic circumstances. The Census Bureau has always been quite clear about the “breaks” in the series and has provided some guidance on how to make more reliable comparisons across time.

Even after accounting for the changes in methodology, it’s pretty clear that the child poverty rate in 2021 still represents a record low. Figure B illustrates SPM child poverty rates as published in blue with the noted series breaks. In red, I’ve constructed an alternative series, in an attempt to “correct” for the series breaks. Census researchers estimated the effect various changes had on the SPM at each break point, and I increased or decreased the historical estimates accordingly to make the SPM child rates more comparable over time. The redesigned questionnaire led to the series break in 2013, while the new processing system led to a series break in 2017. A smaller break in 2019 reflects the implementation of revised methodology. While much ado was made of the need to account for the processing break, which would artificially suggest a drop in the series, other changes moved in the opposite direction. Evidently, the series breaks appear to have little-to-no effect on the conclusion that child poverty is at a historic low. The story is about the drop in poverty over the last two years because of expanded government programs.

Figure B

Child poverty rates fall dramatically with pandemic safety net measures: Child poverty rates using the Supplemental Poverty Measure, as reported and imputed to correct for series discontinuities, 2009–2021

SPM (imputed) SPM (as reported) SPM (as reported) SPM (as reported) SPM (as reported)
2009 17.4 17.0
2010 18.3 17.9
2011 18.4 18.0
2012 18.4 18.0
2013 16.8 16.4 18.1
2014 15.8 17.1
2015 14.9 16.2
2016 13.9 15.2
2017 14.3 15.6 14.2
2018 13.8 13.7
2019 12.6 12.5 12.6
2020 9.7 9.7
2021 5.2 5.2
ChartData Download data

The data below can be saved or copied directly into Excel.

Notes: Using methods outlined in Rothbaum and Edwards 2019, I impute across back in time from the series breaks in 2013 and before for redesigned questions, in 2017 and before for the updated processing system, and in 2019 and before for the implementation of the revised Supplemental Poverty Measure methodology. No adjustment was necessary for the 2020 census-based population controls as Table C-3 shows, there was no statistical difference in the estimation of SPM child poverty with the 2010 and 2020 Census-based controls.

Source: EPI analysis of United States Census Bureau Supplemental Poverty Measure data. 

Copy the code below to embed this chart on your website.

Figure C replicates the exercise from Figure B, imputing child poverty rates by race and ethnicity across breaks in the series. For the original Census data, please compare my figure to Figure 8 of the latest Census Bureau publication on poverty released last week. None of the conclusions made about historically low poverty rates are overturned by these adjustments.

Figure C

Black and Hispanic children experienced a sharp drop in poverty in the last year: Imputed child poverty rates by race and ethnicity, 2009–2021

White, non-Hispanic Black Asian Hispanic
2009 11.0 22.2 17.2 29.4
2010 11.2 25.5 13.9 30.5
2011 11.3 26.2 13.4 30.5
2012 10.8 25.8 16.4 30.5
2013 9.9 24.7 12.8 27.5
2014 8.5 25.2 14.1 26.6
2015 9.1 23.0 12.1 23.5
2016 7.4 23.3 11.0 23.0
2017 8.1 23.7 12.7 22.2
2018 7.0 24.4 11.7 21.9
2019 7.0 20.4 9.5 20.3
2020 5.7 16.9 6.7 14.7
2021 2.7 8.1 5.1 8.4
ChartData Download data

The data below can be saved or copied directly into Excel.

Notes: Using methods outlined in Rothbaum and Edwards 2019, I impute across back in time from the series breaks in 2013 and before for redesigned questions, in 2017 and before for the updated processing system, and in 2019 and before for the implementation of the revised Supplemental Poverty Measure methodology. No adjustment was necessary for the 2020 census-based population controls as Table C-3 shows, there was no statistical difference in the estimation of SPM child poverty with the 2010 and 2020 Census-based controls. Federal surveys give respondents the option of reporting more than one race. In this case the U.S. Census Bureau uses the race-alone concept where, for example, a group such as Asian may be defined as those who reported Asian and no other race. 

Source: EPI analysis of United States Census Bureau Supplemental Poverty Measure data. 

Copy the code below to embed this chart on your website.

There may still be lingering concerns about potential sampling bias before and during the pandemic or the changes to the population controls based on the 2020 decennial census. I’m not terribly concerned that those change the overall story. Below, I explain why.

Sampling bias is certainly a concern, not only during the pandemic when households may have been harder to reach, but even before. Census Bureau researchers have written extensively on the sampling issues before and since the pandemic. The trends in monthly response rates, as shown in Figure 1 here, are troubling; not only did response rates decline, but they differed dramatically between respondents by earnings and income. This has the potential to produce misleading estimates on the state of economic well-being overall and for specific population groups.

Before the pandemic, sampling bias did not lead to statistically significant differences in SPM poverty rates, overall or by race and ethnicity. The drop-off in response rates during the pandemic, particularly when the data were collected for the 2019 data year, did have an impact on poverty rates. Rothbaum and Bee (2022) find the following: “From 2020 to 2022, official poverty rates with the alternative weights were slightly—yet statistically significantly—higher than the survey estimates, by a half percentage point, four-tenths of a point, and three-tenths of a point, respectively.” While they do not provide data on the child SPM rates, or even SPM rates in general, if we were to extrapolate that measurement bias to the SPM imputations in Figures B and C, the results hold. That is, child poverty would still be the lowest on record. It’s also worth noting that the Census-based population controls had no statistically significant impact on SPM child poverty.

All in all, child poverty was the lowest on record in 2021, largely due to key expansions of the Child Tax Credit. Unfortunately, that’s where the good news ends—policymakers let those vital tax credits expire at the end of 2021, leading to higher child poverty rates in 2022.